The aim of this study was analyze modes of mechanical failure in a consecutive cohort of patients and establish possible risk factors. This was a retrospective cohort study. The study was conducted at an academic level-1 trauma center. Twenty-nine patients (mean age 56 years, range 21-92; 45% males, 41% smokers, 17% diabetic, mean body mass index 26.9, range 20-56) with 30 OTA 31A3 fractures treated between 2003 and 2007 were included. Operative fixation using 4.5-mm locking compression plate (LCP) proximal femur plate (Synthes, Paoli, PA). Mechanical failure was defined as loss of alignment of at least 10° or shortening of at least 2 cm. Secondary outcomes included patient and fixation construct variables as possible predictors for mechanical failure. At 20 months of follow-up, 11 failures (37%) occurred. Mean time to failure was 18 weeks (range 2-84). Cumulative failure rates were 10%, 20%, 27%, and 33% at 1, 2, 6, and 12 months, respectively. The most frequent failure mode was varus collapse with screw cut out (5 cases). There was no statistically significant difference between groups with regards to age, body mass index, diabetes, or smoking habit. The presence of a "kickstand screw" and medial cortical reduction were not significantly different in cases with and without failure. Proximal screw number and type were similar in both groups. A high rate of mechanical failure can be expected with proximal locking plate fixation of unstable proximal femur fractures. The use of a "kickstand" screw could not be established to reduce the risk for mechanical failure. Therapeutic Level IV. See Instructions for Authors for a complete description of levels of evidence.