ABSTRACTAlthough the polycross is a useful and cost effective mating design, a lack of paternal pedigree information is a major limitation for polycross breeding in forage grasses such as timothy (Phleum pratense L.). This study describes a paternity test for use in timothy breeding using polymorphic data on 27 genomic simple sequence repeat markers. The paternity test is a simple exclusion statistical test with a combination of maternal information. It successfully determined paternity (success rate = 97%) for 112 progeny plants derived from three polycross groups (A, B, and D). Indirectly selected paternal parents in polycrosses were inferior to maternal parents directly selected by polycross progeny tests mainly for forage yield. Chi‐squared values (χ2) in goodness‐of‐fit tests of the frequency distribution of paternal parents compared with the expected probabilities revealed unbalanced selection in Polycrosses B and D (χ2 = 141.4∗** and 82.7∗**, respectively). Significant differences among the maternal and paternal parents in breeding values for competitiveness toward legumes and low‐digestibility fiber content indicate that unbalanced paternal selection would result from individual phenotypic selection for these traits. These results demonstrate that implementation of a marker‐based paternity test in timothy polycross breeding could significantly improve the selection of superior paternal parents and redress problems of parental imbalance.