[en] The simplification of recording procedures constitutes an economic alternative to reduce operational costs when national recording programs are concerned in low input production systems. A total of 54 940 daily milk yields from a herd of 63 Holstein cows recorded in 3 years were used in this study. The effect of reducing tests to measure the accuracy of 305 milk yields was investigated. Two methods of total milk yield computation (the Test Interval Method (TIM) and the Interpolation using Standard Lactation Curves (ISLC)) were compared based on 495 combinations of milk recording alternatives. These alternatives were based on 2 milkings/day, one alternated milking (AM/PM) and one non-alternated milking (AM or PM). All of them were used with 9 intervals between consecutive tests (3 to 11 weeks) and 11 intervals between calving date and the first test day. Results showed that the differences between the actual 305 d milk yield and milk estimated by TIM were the smallest on consecutive test days of maximum 6 weeks of intervals. Differences increased in both sides positively back to A3 and negatively toward A11. Standard estimation errors increased from 4 to 9 from A3 to A11. Phenotypic correlations decreased from .99 to .94 when intervals between the calving date and the first test day increased and this for all test methods (A3-A 11). The ISLC led to a better accuracy of milk yield estimation when intervals between consecutive tests got longer.
