The efficiency of selection for yield and fruit size using parts of the complete-season record was evaluated using data collected for individual strawberry seedlings (Fragaria x annanasa) in four genetic tests. Part-records were generated for each trait using cumulative weekly sums, and these were compared with complete records to assess the value of extended data collection. In addition, part-records were generated using data from alternate weeks of harvest. Scoring separate sets of seedlings in alternate weeks was assumed to allow a doubling of the unselected population size, halving the selection proportion and increasing selection intensity. Heritabilities estimated for part-records were often larger for mid-season cumulative part-records than for the complete-season records. Seasonal patterns for broad-sense and narrow-sense heritabilities differed, suggesting temporal expression of genes that condition yield and fruit size. Predicted genetic gains were largest using cumulative part-records between 4 and 12 weeks of the 15-week season, depending on trait and breeding objective (breeding value versus clonal value), with 8-9 weeks as a reasonable compromise. Selection using alternate-weeks records was identified as an efficient alternative to selection using total complete-season records. Superiority of the alternate-weeks option depends on the balance between increased selection intensity and decreased individual genetic information, and will differ for specific traits and program designs.