This study attempted to analyze the effect of several factors on the stillbirth pattern in a relatively isolated rural population, La Alpujarra (Spain), during the first half of the 20th century. The study was a retrospective analysis from a total sample of 2199 births to 525 mothers, allowing for birth year of mother, maternal age, parental inbreeding, family size, birth order, sex, single/twin delivery, and birth interval. Binomial probability distribution of stillbirths provided no evidence for any significantly increased risk in relation to family size. Analysis of covariance (ANCOVA) of stillbirth risk in affected families indicated a significant effect for sex of the child, parental consanguinity, and birth year of mother. Logistic regression showed increased risk in twin delivery and pregnancy order one, but not for birth order other than one. Multivariate analysis of variance (MANOVA) testing for differences between affected and unaffected families supported a temporal decrease of stillbirths during the period studied. Although the birth interval average was significantly shorter in affected families (p < 0.0001), this association did not hold, in a more detailed analysis, for individual intervals in these families (p = 0.20). There was no significant effect of maternal age on stillbirths in the whole sample or limited to first pregnancies. These results suggest that birth order one and twin delivery were the main determinants of the stillbirth pattern in La Alpujarra. Furthermore, our data indicate that the decline in stillbirth rate began before medical facilities for perinatal care became available, which was not until after 1950. The temporal decrease in stillbirth rates may therefore be related to an increasing social attention to deliveries rather than to prenatal care medical facilities.