Written and maintained by CASRAI Editorial Board
Last updated
STORM (stochastic optical reconstruction microscopy) builds an image from the positions of thousands of individual fluorescent molecules, each switched on briefly and located one at a time, instead of from a single exposure. The original 2006 report described about 20 nm imaging resolution, roughly ten times better than conventional fluorescence microscopy. What decides the resolution you actually get is the dye, the buffer and the labelling density, which is the focus of this guide.
Short answer
STORM is a single-molecule localization method. In each imaging cycle only a small fraction of the fluorophores in the field of view are switched on, so each glowing molecule is isolated from its neighbours and its centre can be calculated precisely. Repeating this over many cycles and overlaying every localization produces the final image (Rust, Bates and Zhuang, 2006). For a wider comparison with SIM and STED, see the guide to choosing between SIM, STED and SMLM.
What STORM is, and how it relates to dSTORM and PALM
The original STORM used a photoswitchable pair of dyes, Cy3 and Cy5. Red light switches Cy5 into a dark state, and green light restores it, with recovery depending on the nearby Cy3 (Rust et al., 2006). A 2010 review in the Journal of Cell Biology separates three related approaches:
- STORM uses photoswitchable dye pairs such as Cy3-Cy5.
- dSTORM switches conventional organic dyes into a long-lived dark state.
- PALM uses photoactivatable fluorescent proteins instead of organic dyes.
(Schermelleh, Heintzmann and Leonhardt, 2010). The localization principle is shared; the difference is the switching chemistry.
How it works in practice
Because only a sparse subset of molecules is on at any moment, each one appears as a separate spot whose centre can be fitted. The 2006 paper reported a localization accuracy of about 8 nm per switch per cycle, and resolved switches 40 nm apart. A STORM image is then a reconstruction: thousands of raw frames, typically 1,000 or more, are acquired and every localization is plotted. The 2010 review puts the acquisition time at seconds to minutes per image and notes that quality depends critically on labelling density.
Dyes, buffers and what limits resolution
The 2011 Nature Methods study by Dempsey and colleagues compared dyes under two common thiol conditions, MEA and beta-mercaptoethanol (bME), both with a glucose oxidase and catalase oxygen scavenger and 10% glucose. Two properties matter: photons detected per switching event (more photons give a tighter position estimate, with precision scaling with the inverse square root of detected photons) and duty cycle (the fraction of time a molecule is on; a lower duty cycle means more molecules can be localized per diffraction-limited area). The authors state that both a high photon number and a low duty cycle are essential for high-resolution STORM images.
| Dye | Photons per switching event (MEA / bME) | Duty cycle (MEA / bME) |
|---|---|---|
| Alexa 647 | 3,823 / 5,202 | 0.0005 / 0.0012 |
| Cy5 | 4,254 / 5,873 | 0.0004 / 0.0007 |
| Atto 647N | 3,254 / 4,433 | 0.0012 / 0.0035 |
Source: Dempsey et al., 2011. In the same study Atto 647 gave 1,526 photons with MEA and 944 with bME, and Atto 655 gave 1,105 and 657. Across all dyes tested, duty cycles ranged from about 0.0001 to 0.04.
A checklist of what limits your image
- Photons per event. Higher counts improve localization precision.
- Duty cycle. It sets how many molecules can be resolved in one area.
- Survival over many switching cycles. Atto 565 gave the highest photon count in the 2011 comparison but poor image quality, because of early bleaching and few switching cycles. Photons alone do not predict image quality.
- Thiol concentration. For MEA the study found an optimum at a low-to-moderate 1-10 mM, compared with 50-200 mM previously recommended.
- Labelling density. The 2010 review stresses that the result critically depends on it.
3D and multicolour STORM
The 2010 review gives lateral resolution of 20-50 nm (around 30 nm in the running text) and axial resolution of about 100 nm with TIRF, and 20-30 nm for 3D-STORM with TIRF. Astigmatism, biplane and double-helix PSF approaches give roughly 50-80 nm axial resolution over a few microns of depth. These are 2010 figures and later work may differ. For colour, the 2011 study demonstrated four-colour STORM using Atto 488, Cy3B, a red dye (Alexa 647, Cy5 or Dyomics 654) and a near-infrared dye (DyLight 750, Cy7 or Alexa 750). The 2010 review lists two simultaneous colours as a typical capability.
Where it matters for labs and facilities
STORM-type imaging is a sample-preparation and reagent decision as much as an instrument one. The dye, the imaging buffer (thiol plus oxygen scavenger) and the density of labelling all appear in the sources above as direct determinants of resolution. The 2010 review suggests the approach may suit smaller or filamentous structures better than dense, bulky ones. Related reading on CASRAI: the optical microscopy resolution limit, TIRF microscopy and photobleaching and phototoxicity in live-cell imaging.
Common misunderstandings
“STORM gives 20 nm resolution.”
That figure is from the 2006 report. The 2010 review quotes 20-50 nm laterally, and the achieved value depends on photons, duty cycle and labelling density in your sample.
“The brightest dye is the best dye.”
Not according to the 2011 comparison: the dye with the highest photon yield gave poor images because it bleached early.
“STORM is a live-cell method.”
The 2006 paper did not demonstrate live-cell imaging and only suggested it might be possible. The 2010 review describes live-cell use as restricted to small subregions or rather slow processes in 2D.
FAQ
What does STORM stand for?
Stochastic optical reconstruction microscopy. “Stochastic” refers to the random switching of individual molecules on and off.
What is the difference between STORM and dSTORM?
STORM originally used photoswitchable dye pairs such as Cy3-Cy5; dSTORM switches conventional organic dyes into a long-lived dark state (Schermelleh et al., 2010).
How long does a STORM image take?
The 2010 review gives seconds to minutes per image, with 1,000 or more raw frames.
Which dyes work for STORM?
Dempsey et al. (2011) evaluated 26 commercially available organic dyes; Alexa 647, Cy5 and Atto 647N are among those with the photon and duty-cycle values above.
Sources
- Rust MJ, Bates M, Zhuang X. Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nature Methods 2006;3(10):793-795. PMC2700296. Checked 2026-10-29.
- Dempsey GT, Vaughan JC, Chen KH, Bates M, Zhuang X. Evaluation of fluorophores for optimal performance in localization-based super-resolution imaging. Nature Methods 2011;8(12):1027-1036. PMC3272503. Checked 2026-10-29.
- Schermelleh L, Heintzmann R, Leonhardt H. A guide to super-resolution fluorescence microscopy. Journal of Cell Biology 2010;190(2):165-175. PMC2918923. Checked 2026-10-29.
