Estrus detection efficacy and heat detection protocol were studied by means of a field study carried out on 878 lactating beef cows in 60 French herds. Average herd size was 48, and 75% of the farmers partly or exclusively used artificial insemination. The cows were calved between October 1992 and March 1993. Estrus was recorded daily by the farmers. Cycling status was determined by progesterone radioimmuno-assay 2 months after calving. The relationship between the estrus detection protocol and the delay period from calving to first observed estrus was analysed using survival curves and the Cox proportional hazard model, adjusting for confounders. Seventy-one percent of the cows were seen in estrus by the farmers; the interval between calving and the first observed estrus ranged from 9 days to more than 5 months and the median was 56 days. Two months after calving, 44% of the cycling cows had not been seen in heat by the farmers and 11% of the non-cycling cows had been reported to have been in estrus. The heat detection protocol varied widely between farmers, depending on the considered estrus signs, schedule and time spent looking for signs. Two factors were significantly related to a shorter interval from calving to first observed heat: the use of artificial insemination (which relates to the farmer's interest in heat detection) and an overall daily time spent for heat detection greater than 1 h. Cows in tie stalls had a delayed interval to the first observed estrus. These results show that many farmers did not adapt their reproduction practice sufficiently to an earlier calving period. There is room for improvement since in many cases the heat detection protocol does not match the required standards for optimal heat detection.