Soil quality of tillage systems receives much attention worldwide, although few studies attempt to link soil quality to yield. Partial least‐squares regression analysis is a suitable method to construct predictive models around plural, highly collinear factors, such as soil quality and its effects on yield. This study aimed at identifying the soil quality properties which best model pasture herbage yield by relating soil quality indicators with variations in yield as a result of soil disturbance caused by tillage. The study was conducted on kikuyu (Pennisetum clandestinum) over‐sown with annual ryegrass (Lolium multiflorum) using different tillage methods. Tillage methods involved over‐sowing kikuyu with ryegrass using a minimum‐tillage seed drill, eradication of kikuyu with herbicide and ryegrass sown with a minimum‐tillage planter, shallow and deep disturbance, and a control. Most changes in soil quality indicators as a result of soil tillage were observed shortly after tillage and occurred mostly at the 0–100 mm soil layer. Few of these effects were still visible 420 d after tillage. Microbiological indicators changed most in response to the treatments, but unlike the chemical and physical indicators, microbiological indicators have no causal link to pasture production. The contribution of individual soil quality indicators to variance in pasture productivity could not be isolated and should thus be seen as complex processes which affect yield. Although mechanisms of how some soil quality indicators affect yield is clear, more research is required to determine mechanisms of how a combination of multiple soil quality indicators affects yield.