Unlike tri-compartmental arthroplasty, unicompartmental knee arthroplasties (UKAs) correct only osteochondral condylar defects and do not include any ligament balancing. Pre-operative deformities of more than 20° strongly suggest that UKA is not indicated. Best results are generally obtained by avoiding hypercorrection and not exceeding 5° of residual deformity post-operatively. Since 1997, 112 UKAs have been implanted. We performed tibial osteotomies with respect to tibial plateau inclination in the frontal plane (metaphyseal axis). Our patients were pre-operatively studied and then re-evaluated after a mean follow-up of 4 years. We used the GIUM (Gruppo Italiano Utilizzatori Monocompartimentali) scoring system. Pre-operative and post-operative radiograms from all patients were collected, and then we correlated the amount of correction of lower limb mechanical axis with GIUM score for each patient. Mean pre-operative mechanical axis of the lower limb showed a varus deformity of 7.43°, whereas post-operative values averaged 5.56° of varus deformity, with a minimal valgus correction of the deformity. Thus, correction of angular deformity was statistically negligible (p> 0.5). Mean pre-operative GIUM score was 20.3, whereas mean post-operative score was 71.6 (p< 0.001). Correlation between entity of correction and GIUM score was significant (r=0.76). The amount of angular correction of the mechanical axis of the lower limb was statistically significant in influencing functional outcomes (p< 0.05). The best results are obtained by correcting excessively valgus knees to a physiologic range, while varus knees have to be minimally corrected. The worst results are obtained with greatest modifications of the mechanical axis of the lower limb.