The focus of this work is the analysis of changes in completed family size and possible determinants of that size over time, in an attempt to characterize the evolution of reproductive patterns during the demographic transition. With this purpose in mind, time trends are studied in relation to the mean number of live births per family (as an indirect measure of fertility), using family reconstitution techniques to trace the reproductive history of each married woman. The population surveyed is a Spanish rural community called Lanciego, located at the southern end of the province of Alava (Basque Country). A total of 24,510 parish records of baptisms, marriages and burials made between 1800 and 1969 were examined to obtain the demographic data set. For each reconstituted family, the variables included in the study were the number of live births per family or family size (FAMS), year of marriage (YEAR), age at marriage of both partners (AMAN, AWOM), wife's age at the end of marriage (WEND), duration of marriage (MARD), age at first maternity (A1CH), length of reproductive span (REPS) and number of children dying before their first anniversary (MINF). Through a principal component analysis, three factors were found that explained more than 75% of the total variance. Association of variables in factors I and III was particularly useful in characterizing the variability of mean family size in pre-transitional, transitional and post-transitional cohorts. During demographic transition, a decreasing trend is observed in the variables FAMS, REPS and MINF, while variables AWOM, AMAN, WEND and A1CH show a tendency to increase over the 20th century. Results obtained by multiple regression analysis confirm that the best predictors of family size (dependent variable) were REPS and MINF, which between them explained over 85% of the total variation in FAMS (R2 = 0.853). In Lanciego, birth control seems to be present on the evidence of an increase in age at first maternity and a decrease in age at last parturition, indicating that the beginning of the reproductive span is delayed and its end is brought forward. Interaction between family size and infant mortality is discussed in the light of various hypotheses, including replacement of descendants, the so-called biological effect and the theory of r and k selection.