Courses on capture-recapture techniques in epidemiology and the social sciences often focus onthe estimation of the population size using the cross tabulation of multiple, incomplete registration files. In this paper we emphasize two aspects of population size estimation that deserve more attention than they usually get. One of these concerns the possibility to obtain a population size estimate from single registration file with zero-truncated count count data. The other one is the use covariates in order correct for heterogeneous capture probabilities and to obtain insight in the composition of the population. This is also possible if the covariates are not measured in each registration file.