Good H-Index

There is no single number that counts as a “good” h-index across the board. The question is only answerable once three things are fixed: field (citation density and publishing pace vary enormously by discipline), career stage (h-index can only grow, so it mechanically rewards seniority over quality), and database (Google Scholar, Scopus and Web of Science index different content and can produce three different numbers for the same person). An h-index that is unremarkable in cell biology can be exceptional in pure mathematics; an h-index that looks modest for a full professor can be outstanding for a fourth-year PhD student; and a number pulled from Google Scholar is not directly comparable to the same person’s Scopus figure. Any claim that gives a bare number as “good” without naming a field, a career stage, and a source database is not measuring what it appears to measure.

Why the raw number can’t be judged on its own

The h-index is defined as the largest number h such that a researcher has h papers each cited at least h times. That construction bakes in three sources of variation that make cross-context comparison unsound:

• Field-dependence. Citation practices differ by discipline — average reference-list length, typical time-to-citation, and the size of the citing community are not comparable between, say, biomedicine and mathematics. Jorge Hirsch acknowledged this in the paper that introduced the metric, noting there will be “differences in typical h values in different fields, determined in part by the average number of references in a paper in the field,” and that h values in the life sciences run much higher than in physics (Hirsch, PNAS 102(46):16569–16572, 2005).
• Career-stage (time) dependence. Because h-index can only increase or stay flat as a career progresses, it partly measures how long someone has been publishing, not just how good the work is. Two researchers with identical h-index values 10 years apart in career length are not equivalently accomplished, and the same mechanism structurally disadvantages anyone with a non-linear career — a career break, part-time research years, or a late start — against a continuously-publishing peer of the same chronological career length.
• Database-dependence. Google Scholar, Scopus and Web of Science index different content and different citation windows, so the same researcher can show three different h-index values depending on which database produced it. A number is not comparable to a benchmark figure computed from a different source. See “Where to find your h-index” below for what actually drives that gap.

Hirsch’s own field-specific benchmark — and why it doesn’t generalize

The closest thing to an authoritative “good h-index” number comes from Hirsch’s original 2005 paper itself, and it is narrower than it is usually quoted as being. Working from typical values he observed among physicists, Hirsch proposed a rough scale by career length:

Career length (active research)	Hirsch’s descriptive tier	Approximate h-index
~20 years	“Successful scientist”	~20
~20 years	“Outstanding scientist, likely to be found only at the top universities”	~40
~20 years	“Truly unique individual”	~60
~30 years	“Truly unique individual”	~90

Source: Hirsch, PNAS 2005. This table is reproduced directly from Hirsch’s own descriptive tiers — it is not a cross-field standard. Two caveats matter more than the numbers themselves:

• Hirsch derived these figures specifically from physics publication and citation patterns and said so explicitly — he did not offer them as a cross-field standard, and noted that further research would be needed to understand how h-index distributions differ across other fields.
• The figures are two decades old. Publication volume, co-authorship norms, and citation databases have all changed since 2005; treat the specific numbers as a historical illustration of how field-specific a “good” h-index claim has to be, not as a current threshold to apply to anyone’s CV.

How much field variation actually looks like

Rather than extrapolate a “good h-index by field” table of absolute numbers from Hirsch’s single, dated, physics-only data point — something a number of researcher-facing blogs do, and which overstates what that one source supports — the more defensible published evidence on cross-field variation is a relative normalization factor, not an absolute threshold. Iglesias and Pecharromán (Scientometrics 73:303–320, 2007) measured mean citations per paper across 21 Web of Science subject fields (1995–2005) and proposed a scaling factor to bring h-index values from different fields onto a common scale, using physics as the reference point.

Field group (illustrative, from Iglesias & Pecharromán 2007)	Relative citation density vs. physics	Effect on raw h-index at equal “true” impact
Mathematics, computer science	Lower (mean citations/paper under ~2.5 in their sample)	Raw h-index tends to run lower than in higher-citation-density fields for comparable impact
Physics (reference field)	Baseline (factor = 1)	—
Molecular biology, biomedicine	Substantially higher (mean citations/paper over ~24 in their sample)	Raw h-index tends to run considerably higher than in physics for comparable impact

This table reports Iglesias & Pecharromán’s published field-level citation-density comparison, not a “good h-index” lookup table — it explains why a given raw number means something different in different fields, it does not hand you a target number. Treat it as directional: it tells you that a raw h-index in a low-citation-density field (mathematics, computer science) that looks small next to one in a high-citation-density field (molecular biology) may represent comparable underlying impact, not less of it. It does not tell you what number is “good” in either field.

The m-quotient: Hirsch’s own attempt at career-stage normalization

In the same paper, Hirsch proposed a second metric specifically to correct for the time-dependence problem: the m-quotient, defined as the h-index divided by the number of years since a researcher’s first published paper (m = h / n). Because raw h-index only grows, Hirsch intended m as “a useful yardstick to compare scientists of different seniority” — an early-career researcher with h = 10 after 5 years (m = 2) and a senior researcher with h = 40 after 20 years (m = 2) are, on this measure, progressing at a comparable rate, despite a fourfold difference in raw h-index.

The m-quotient inherits h-index’s field-dependence (it normalizes for time, not discipline) and is unstable early in a career, where small changes in h produce large swings in m. It is a partial fix for one of the three problems above, not a solution to all of them, and it is far less commonly reported than the raw h-index itself. It also does not correct for career breaks or part-time research appointments — “years since first paper” keeps counting through parental leave, illness, part-time contracts, or a period spent outside research, which mechanically lowers m for researchers with non-continuous careers relative to continuously full-time peers of the same nominal seniority.

Is a higher h-index always better? Why “highest h-index” is the wrong question

Searches for the “highest h-index” usually want a ranking of specific individuals. CASRAI is not publishing one, for the same reason the bibliometrics literature treats the raw number with caution even at the top of the distribution: a higher h-index is not straightforwardly “more impact,” and the metric has documented failure modes that get worse, not better, at scale.

• It is insensitive to outstanding individual papers. Once a paper has accumulated h citations, any further citations to that paper do nothing to raise the h-index — a single paper cited 50 times counts identically to one cited 5,000 times, provided both already clear the h threshold. Waltman and van Eck (“The inconsistency of the h-index,” Journal of the American Society for Information Science and Technology 63(2):406–415, 2012) formalized this as an inconsistency in how the h-index ranks scientists, showing it can produce rankings that reverse under seemingly minor changes to the underlying publication set.
• It can be inflated through self-citation. Bartneck and Kokkelmans (“Detecting h-index manipulation through self-citation analysis,” Scientometrics 87(1):85–98, 2011) modeled how strategically placed self-citations can raise a researcher’s h-index and proposed a detection metric (the q-index) precisely because the manipulation is measurable and real, not hypothetical. Citation cartels — coordinated reciprocal citation among a group of authors — work the same way at group scale.
• It rewards volume and seniority over selectivity. As covered above, h-index can only rise, so a long career with a high publication rate mechanically outproduces a shorter or more selective one on this metric alone, independent of the quality difference between the two bodies of work.

The honest answer to “what’s the highest h-index” is: it is not a meaningful question on its own, for the same reasons “what’s a good h-index” isn’t — the number at the top of any cross-field, cross-database ranking reflects field, career length, database coverage and citation behavior at least as much as it reflects research quality.

What responsible-assessment frameworks actually say

Several widely adopted frameworks address single-number research metrics generally, and h-index specifically falls under their scope even where it isn’t named:

• The Leiden Manifesto (Hicks et al., Nature 520:429–431, 2015) states as one of its ten principles that assessment should “account for variation by field in publication and citation practices,” and separately warns against “misplaced concreteness and false precision” — directly applicable to treating an h-index difference of one or two points as meaningful.
• DORA (the San Francisco Declaration on Research Assessment) is framed principally around journal-level metrics such as the Journal Impact Factor; its core recommendation not to use a journal-based metric “as a surrogate measure of the quality of individual research articles, to assess an individual scientist’s contributions, or in hiring, promotion, or funding decisions” mentions h-index only once, in passing, as one of several alternative journal-level indicators. DORA’s general caution against reducing research quality to a single number is nonetheless the principle that subsequent responsible-metrics guidance extends explicitly to researcher-level indices like h-index.
• The CoARA Agreement and the Hong Kong Principles both commit signing institutions to basing assessment on qualitative expert judgement of a researcher’s actual contributions rather than on standalone bibliometric thresholds.

None of these frameworks publish a numeric “good h-index” figure, by field or otherwise — that omission is itself part of their point: a defensible research-assessment process does not reduce to a lookup table.

Where to find your h-index

Because the number is database-dependent, the practical first step is usually not “what’s a good h-index” but “which h-index, from which source.” The three sources researchers most commonly cite:

• Google Scholar. A Google Scholar Citations profile computes and displays an h-index automatically on the author’s public profile page, alongside a separate i10-index. Google Scholar tends to index a broader, less curated set of sources — including preprints, theses, conference proceedings and grey literature — than the other two, which typically pushes its h-index figure higher than the equivalent Scopus or Web of Science number for the same person.
• Scopus. A researcher’s Scopus Author Profile (found via Scopus Author Search, keyed to a Scopus Author ID) reports an h-index computed only from Scopus-indexed publications and their Scopus-indexed citations — a narrower, more curated source base than Google Scholar. See Google Scholar vs. Scopus for how the two databases’ coverage differs more generally.
• Web of Science. A Web of Science Researcher Profile (often linked to a ResearcherID) reports an h-index computed from Web of Science Core Collection content, historically the most conservatively curated of the three source bases.

Because coverage differs, an h-index from any one of these is only meaningful when the source and pull date are stated alongside it — and only comparable to another figure pulled from the same database.

What to do instead of looking for a number

• Compare within field and career stage, not across them. If a comparison is needed at all, benchmark against researchers in the same or an adjacent subfield with a similar number of years since their first publication or since PhD completion — not against a discipline-wide or cross-field figure.
• Hold the database constant. Don’t compare an h-index pulled from Google Scholar against one pulled from Scopus or Web of Science; state which database and date the figure came from.
• Use it as one input among several, not a threshold. Per the Leiden Manifesto and CoARA, quantitative indicators should support, not replace, qualitative review of the actual body of work — the papers themselves, their contribution, and context such as career breaks, part-time research roles, or a switch between subfields.
• Watch for self-citation and reciprocal-citation patterns when a reported h-index looks unusually high relative to a researcher’s field and career length — it is a documented, measurable manipulation vector, not just a theoretical risk.
• Be skeptical of any “good h-index by field/career stage” table you encounter, including informal ones from other websites, unless it discloses its data source, database, sample and date. Absent that disclosure, the number is not verifiable and is not more reliable than an average of anecdote.

References

• Hirsch JE. “An index to quantify an individual’s scientific research output.” PNAS 102(46):16569–16572, 2005.
• Iglesias JE, Pecharromán C. “Scaling the h-index for different scientific ISI fields.” Scientometrics 73:303–320, 2007 (doi:10.1007/s11192-007-1805-x).
• Waltman L, van Eck NJ. “The inconsistency of the h-index.” Journal of the American Society for Information Science and Technology 63(2):406–415, 2012.
• Bartneck C, Kokkelmans S. “Detecting h-index manipulation through self-citation analysis.” Scientometrics 87(1):85–98, 2011.
• Hicks D, Wouters P, Waltman L, de Rijcke S, Rafols I. “Bibliometrics: The Leiden Manifesto for research metrics.” Nature 520:429–431, 2015 (doi:10.1038/520429a).
• San Francisco Declaration on Research Assessment (DORA), sfdora.org.
• Coalition for Advancing Research Assessment (CoARA) Agreement, coara.eu.
• Hong Kong Principles for Assessing Researchers, 2020.