Fecundity, an important demographic variable for wildlife populations, is time consuming and expensive to measure. For seabirds, reproductive success is often estimated from 2 surveys of the colony, 1 during incubation and 1 during chick rearing. Using 33 yr of data on Magellanic Penguins (Spheniscus magellanicus), we compared reproductive success estimated from 2 surveys per season, from infrequent nest checks, and from daily nest checks to answer 4 questions: (1) Is reproductive success estimated from 2 surveys per season accurate enough for population models (within 15% of the ‘true' value)? We found that the answer was yes, if surveys were done on optimal dates and chick counts were adjusted by the percentage of chicks that were heavy enough to fledge. Optimal dates depended on the synchrony and phenology of the species. If penguins bred particularly late or early in a given year, estimates were wrong even if counts were done on the average optimal date. (2) Can counts of occupied nests be adjusted based solely on date? We determined no, not without additional information on egg dates and egg and chick losses. (3) Can long-term trends in reproductive success be detected from 2 surveys per season? Our answer was yes; even using biased data from counts on suboptimal dates, trends with similar slopes were detected using both methods (2 surveys vs. nest checks). (4) How often must nests be checked to get reproductive success estimates comparable with those calculated from daily nest checks? We discovered that checking nests every 2 days was as good as doing daily checks. Reproductive success was overestimated by <5% when checks were conducted every 3–8 days and by <15% (in 90% of years) when checks were done every 9–30 days. The degree of overestimation depended on the timing of nest checks relative to egg laying, loss, and hatching. We conclude that reproductive success can be estimated from 2 surveys, but only if timing, synchrony, and variability of breeding are also known.