Abstract: Forty weanling Quarter Horses, from the Pitchfork Ranch, Dickens, Texas, were randomly assigned to one of five groups based on the amount of handling they would receive. The five groups were: (1) limited handling; (2) 1 week of handling; (3) 2 weeks of handling; (4) 3 weeks of handling; (5) continuous handling for 18 months. As 2-year-olds, the horses were tested in a simple place-learning T-maze after being preconditioned. A maximum of 30 daily trials were conducted for 20 consecutive days, with feed location alternating between sides on successive days. If a horse met the criterion of 11 out of 12 correct responses with 8 correct responses being consecutive, it was retired without completing the 30 trials. Group 1 made fewer correct responses during the 20 days than any other group except Group 3. All groups achieved learning by Day 10, but the most-handled group reached a consistently higher percentage of correct responses earlier than the less-handled groups. Group 5, the group receiving the most handling, was the least emotional, as determined by the animals' reactivity to a novel stimulus, and received the highest scores for trainability after being ridden. Preconditioning trainability scores, or estimates of the horses' ability to learn prior to conditioning and testing, tended to predict maze-learning ability. These scores were also significantly correlated with post-maze training scores.