Objective. To provide a best estimate of the average treatment effect when microfracture was chosen as the intervention of choice in patients with full-thickness cartilage defects of the knee. Design. We focussed on controlled studies which either referred to microfracture alone or in comparison with any other surgical treatment of articular cartilage of the knee. Papers including patients who had been treated by microfracture and concomitant adjuvant procedures like ACL reconstruction or meniscus repair were accepted too, whereas papers reporting on the microfracture technique combined with the implantation of a scaffold were excluded. To achieve a best estimate of the average, to be expected treatment effect we pooled pooled before–after data of study arms using microfracture. Because cartilage studies employ various scales to measure functional improvements, we standardized treatment effects using Hedges' g. To provide clinically meaningful estimates we converted the pooled summary effect back into the respective scales by multiplying the pooled effect with pooled standard deviations of each included clinical scale. Results. A systematic review of the literature revealed six papers including 200 patients with a mean age of 32 years, a mean defect size of 3 cm. 2. and a follow up period from 2 to 5 years. Four of the studies compared microfracture to autologous chondrocyte implantation and two of them to osteochondral autologous transplantation. All patients were treated by the microfracture technique as described by Steadman and by a similar rehabilitation protocol which only allowed crutch-assisted touchdown weight bearing initially. Referring to the individual studies, a comparison of the pooled estimates of Hedges' g revealed that the two papers which evaluated the youngest patients provided the highest treatment effect. On the contrary, those two papers which focussed on the largest lesions, reported the worst improvement. Finally, the remaining two papers whose patients were characterized by similar age and defect size presented comparable results. The individual standardized effect sizes were combined into an overall best estimate. Its value was 1.678, measured in units of standard deviation, with the 95% confidence interval of [1.016; 2.340] resulting in different values of the average, to be expected treatment effect when it is measured in Lysholm Score (22.1), IKDC Score (26.5) and KOOS (15.2) points. Conclusions. Our results offer a clinically intuitive estimation of the average treatment effects on common clinical scales. Compared to the preoperative situation, a significant clinical improvement can be expected for each patient. Nevertheless, the magnitudes of these treatment effects are an approximation and must be interpreted cautiously. Furthermore, we did not succeed to confirm that young age and small lesion size have a beneficial effect on the clinical outcome