Many workforce development training programmes use learning gains as a measure of programme effectiveness. However, research on K-12 education suggests that posttest scores may be influenced by pretesting effects. Pretesting may improve posttest performance by giving learners preknowledge of posttest content. Alternatively, pretesting may enhance learning via the retrieval practice effect. Cross-classified analysis was used to detect pretesting effects in archival data from a large-scale agriculture workforce development programme. Data included scores from 14,302 learners, 352 trainers and 261 training centres. Results indicated that pretesting led to higher scores on posttests. Follow-up logistic regression analyses found no effect for time lag between pretest and posttest. In addition, pretested learners did not perform better on items that were included on both forms, suggesting that score differences were due to retrieval practice effects rather than preknowledge of posttest content. These results provide evidence that pretesting may enhance learning in large-scale workforce development training courses.