Abstract Objective Trunk training after stroke is an effective method for improving mobility, yet underlying associations leading to the observed mobility carryover effects are unknown. The purposes of this study were to investigate the effectiveness of trunk training for gait and trunk kinematics and to find explanatory variables for the mobility carryover effects. Methods This study was an assessor-masked, randomized controlled trial. Participants received either additional trunk training (n = 19) or cognitive training (n = 20) after subacute stroke. Outcome measures were the Tinetti Performance-Oriented Mobility Assessment (POMA), the Trunk Impairment Scale, spatiotemporal gait parameters, center-of-mass excursions, and trunk and lower limb kinematics during walking. Multivariate analysis with post hoc analysis was performed to observe treatment effects. Correlation and an exploratory regression analysis were used to examine associations with the mobility carryover effects. Results Significant improvements after trunk training, compared with the findings for the control group, were found for the Trunk Impairment Scale, Tinetti POMA, walking speed, step length, step width, horizontal/vertical center-of-mass excursions, and trunk kinematics. No significant differences were observed in lower limb kinematics. Anteroposterior excursions of the trunk were associated with 30% of the variability in the mobility carryover effects. Conclusions Carryover effects of trunk control were present during ambulation. Decreased anteroposterior movements of the thorax were the main variable explaining higher scores on the Tinetti POMA Gait subscale. However, the implementation and generalizability of this treatment approach in a clinical setting are laborious and limited, necessitating further research. Impact Trunk training is an effective strategy for improving mobility after stroke. Regaining trunk control should be considered an important treatment goal early after stroke to adequately prepare patients for walking.