Obtaining an accurate quantification of population size is often of prime importance in ecology and conservation biology (e.g. population viability analysis, a basic step for assessing species and population status in a given area and guiding effective conservation). When obtaining a reliable quantification of absolute (vs. relative) population size is required, Mark–Release–Recapture (MRR) is a method of choice for many organisms. This is a highly reliable but costly procedure in terms of time and potential impact on species and sites. Consequently, less costly alternatives are highly desirable for conservation and population ecologists. We present here a simplified MRR protocol to mitigate this cost of repeated MRR sampling with little compromise on the quality of the population size estimation. Using one of the largest existing butterfly MRR databases, collected on two fritillary species over a period of 20 years and >20 populations in Belgium, we assessed the possibility to reduce the effort of collecting MRR data while keeping accurate quantification of total population size. By downsampling from the full datasets and calculating a range of demographic census metrics, we specifically investigated whether marking individuals is necessary, and whether the number of sampling sessions can be reduced. We found that counting individuals is not enough: some individual marking, even in a simplistic way to differentiate newly recorded from previously seen individuals, is essential for estimating population size. A simple linear conversion function (number of “missed” individuals for each marked one) can be used to compute population size from the number of individuals marked over a small number of MRR sampling sessions. Parameterizing this function is system specific, because it depends on detectability of individuals, but only requires an initial effort of traditional high‐effort MRR in a few populations encompassing the expected range of population size, combined with previous knowledge on the species about potential factors affecting detectability. Our simplified MRR protocol should allow scientists to obtain absolute population size estimates almost as good as with traditional high‐effort MRR, but at a cost lowered in both the marking procedure and the intensity of field visits.