Estimates of periphyton abundance, biomass (as biovolume, chlorophyll, dry mass, or ash-free dry mass), and production were reanalyzed to examine the effect of the mean value, sampler size, type of substratum, and habitat on the variance of these estimates. There were strong relationships between the variance and the mean for all variables considered. Variability per sampling unit increased with sampler size for all variables except production and biovolume, which were both unaffected. Chlorophyll estimates were significantly less variable on artificial than on natural substrata, but this trend was not confirmed by other variables. For biovolume, the trend was reversed. Habitat affects periphyton variability only for some variables: production was more variable in streams than in lakes whereas chlorophyll was less variable in artificial streams than in any other habitat. Although periphyton tends to be less patchy than benthos, detection of small differences still requires a very intense sampling effort. Since the median replication is 3–10 in most published studies, only large differences can be detected with 95% probability, from twofold for the most precise variable (biovolume) to eightfold for the less precise variable (dry mass). Our ability to detect patterns in periphyton ecology is thus very limited.