Appropriate survey design is essential to estimate population size reliably, especially for endangered species. Dupont's Lark Chersophilus duponti, one of the most endangered passerines in Europe, has been monitored using diverse counting methods. This variation in the methods employed may have a significant effect on the estimates of population sizes. The present study compares four methodologies cited in the literature as having been used for Dupont's Lark censuses: the Finnish line transect (25 m inner belt width), line transect (500 m inner belt width, a specific application of the transect method developed for the second national census of the species in Spain), mapping and point counts methods. We also determined the adequate number of visits needed to detect a reliable number of territories by the mapping method and analysed the effect of census month on the number of males recorded in line transects. According to our results, we consider that the mapping method, based on four visits, is the most adequate methodology for monitoring the species. However, for surveys in large areas the use of the line transect method during May and June may be more affordable. Although census date did not have a significant effect on the number of males recorded by line transect surveys, a larger number of males were indeed detected during May and June. Thus, censuses carried out in March may lead to greater underestimation of population sizes. Nonetheless, the best date for counting Dupont's Lark may differ for populations located at lower altitude. Point counts and Finnish line transects should not be used for counting the species, since both overestimate population sizes by about 35%, according to the mapping method and line transect results.