Distributions of microphytobenthos are highly heterogeneous at scales as small as a few centimetres. However, the sampling protocols currently used for the absolute determination of microphytobenthos biomass through chlorophyll a concentration measurements in surface sediments are too limited to take this variability into account, typically relying on 3-5 samples taken within a randomly located 1 m2 quadrat in a given environment. We address this issue by objectively and quantitatively inferring the minimum number of samples required to obtain reliable estimates of microphytobenthos biomass on the basis of high-resolution sub-sampling (225 regularly spaced samples) within each of nine 1 m2 quadrats at an unvegetated sheltered intertidal sandbank of the Gold Coast (Queensland, Australia). The results were generalised using data obtained in previous studies on an exposed sandy shore and on sheltered estuarine sandy muds of the Eastern English Channel. Estimates of chlorophyll a concentration exhibited a high degree of heterogeneity, both between and within quadrats. The number of samples needed to estimate the average chlorophyll a concentration, and hence mean microphytobenthos biomass with 95% confidence intervals, ranged from 15 to 115, and mainly depended on the presence of global and local gradients within the quadrats. These results have major implications for intertidal ecology by implying a possible systematic bias in the measurement of both microphytobenthos biomass and production of up to 40%. Finally, we emphasise that this issue can be circumvented using field spectrometry or PAM fluorescence measurements coupled with traditional sediment sampling techniques, and urge for unified protocols to be adopted for the routine use of these combined methods.