With a rise in concern for animal welfare, the equine world has started using positive reinforcement (R+); as such, horses often experience a combination of negative reinforcement (R-) and R+. I compared the effects of R- to a combination of positive and negative reinforcement (R-/R+) training. Horses were trained to walk across two visually discriminable liverpools (striped, Experiment 1; colored water, Experiment 2), each associated with either R- or R-/R+, and training type alternating across six days. I measured highest training criteria reached, prevalence of undesirable behaviors, salivary cortisol (pre- and post-training), time spent by the trainer in motionless human tests (pre- and post-training), and horses' preference for the two liverpools using concurrent choice. Across both experiments, I found no significant difference in the proportions of criteria reached between training types; horses engaged in mugging for longer periods of time in R-/R+ than R-; no significant difference between training types for the pre- to post-change of cortisol; a greater proportion of horses increased time spent with R-/R+ trainer than the R- trainer; and no difference between first choice in the preference test or time horses spent in proximity to the liverpool, based on the training type with which the liverpool was associated. Overall, I found few differences between R-/R+ and R-, which could be due to horses only having 30 min total training contact with either training, or my use of relatively low intensities of R- and R+. / Master of Science / The equine world has started using positive reinforcement (R+), such as providing treats. Often horses experience a combination of negative reinforcement (R-) and R+, such as having rein pressure released and being given a treat. I compared effects of R- to a combination of positive and negative reinforcement (R-/R+) training. Horses were trained to walk across two visually distinct liverpools, a 1 m X 2.7 m shallow pool, (striped, Experiment 1; colored water, Experiment 2) each associated with either R- or R-/R+, and training type alternating across six days. I measured highest training level reached, occurrence of undesirable behaviors, salivary cortisol (a measure of stress), time spent by the trainer in motionless human tests, and horses' preference for the two liverpools. Across both experiments, I found no significant difference in the proportions of criteria reached between training types; horses investigated the trainer for treats for longer durations in R-/R+ than R-; no significant difference between training types for the pre- to post-change; a greater proportion of horses increased time spent with R-/R+ trainer than R- trainer; and no difference between first choice in the preference test or time horses spent in proximity to the liverpool, based on the training type with which the liverpool was associated. Overall, I found few differences between R-/R+ and R-, which could be due to horses only having 30 min total training contact with either training, or my use of relatively low intensities of R- and R+.