The aim of this study was to compare different statistical methods for the estimation of daily and 305-day lactation milk, fat and protein yields of Holstein and Simmental cattle breeds using an alternative milk recording scheme. Data included 6,824 individual test-day milk yield records collected according to the A4 milk recording method on 668 cows reared on 15 family farms. Daily milk, fat and protein yields were estimated using several statistical methods with regard to breed. The 305-day lactation yields were calculated from estimated daily yields using the Test Interval Method. The correlation between estimated and true yields, as well as the mean difference among estimated and true yield were used as the evaluation criteria for estimation methods. The linear regression of daily to partial milk, fat and protein yields while taking into account the interval between successive milkings was shown to be the most accurate model for estimating daily values, either from morning or evening records. The simple doubling of morning or evening records overestimated and underestimated the daily yields, respectively. When 305-day lactation milk, fat and protein yields were compared no difference between evaluated methods were found. Also, a separate estimation of daily and 305-day lactation yields according to breeds did not result in increased estimation accuracy.