The active germplasm Bank of the Integrated Unit (Facultad de Ciencias Agrarias, Unversidad Nacional de Mar del Plata and the Experimental Station of INTA) at Balcarce, keeps both in vivo (live herd) and in vitro Criollo Cattle (samples of semen and embryos cryogenically), with the aim of preserving genetic variability for potential future use. Parameters based on probable origin of genes, on population size and on the estimation of components of (co)variances for birth and weaning weights were estimated to evaluate the genetic variability of the material contained in the Bank. The information available came from 667 individuals, of which 84 had unknown parents, and 142 half-founders (only one parent known), having a mean number of births per year of 24 calves from four different sires. AI was performed during the mating season almost every year. The evolution of effective number of founders (fe), effective number of ancestors (fa), effective number of founder genomes (Ng) and the ratio fe/fa over different periods from the creation of the Bank, were analysed. Annual increase of inbreeding coefficient (F) was 0·2% and the effective population size (Ne) was estimated as 23, 14 and 25 individuals for the periods defined by years 1978–82, 1983–87 and 1993–97, respectively. The ratio fe/fa was higher in the years where the ratio number of sires/number of dams was lower, indicating that the impact of the effective number of sires is more important than the effective number of females in generating genetic bottlenecks. When the reference population was taken for the period 1989–97, 16·45 out of the 85·63 founders of the population made contributions not explained by other ancestors, but if possible losses of genomes due to genetic drift are considered, the number of genomes still present was reduced to 9·09 indicating a possible narrow genetic bottleneck in the last period. The Ng/fe ratio was 0·11, 0·10 and 0·11 for the reference populations defined by years 1989–92, 1993–97 and 1989–97, respectively. The main objective in germplasm conservation programmes should be the maintenance of genetic variability. Thus, matings in the population under conservation should be designed in order to mate those animals that maximize the effective number of genome founders (Ng) to avoid losses due to genetic drift.