Increasing numbers of clinical trials have demonstrated the benefits of incomplete caries removal, in particular in the treatment of deep caries. This study systematically reviewed randomized controlled trials investigating one- or two-step incomplete compared with complete caries removal. Studies treating primary and permanent teeth with primary caries lesions requiring a restoration were analyzed. The following primary and secondary outcomes were investigated: risk of pulpal exposure, post-operative pulpal symptoms, overall failure, and caries progression. Electronic databases were screened for studies from 1967 to 2012. Cross-referencing was used to identify further articles. Odds ratios (OR) as effect estimates were calculated in a random-effects model. From 364 screened articles, 10 studies representing 1,257 patients were included. Meta-analysis showed risk reduction for both pulpal exposure (OR [95% CI] 0.31 [0.19-0.49]) and pulpal symptoms (OR 0.58 [0.31-1.10]) for teeth treated with one- or two-step incomplete excavation. Risk of failure seemed to be similar for both complete and incomplete excavation, but data for this outcome were of limited quality and inconclusive (OR 0.97 [0.64-1.46]). Based on reviewed studies, incomplete caries removal seems advantageous compared with complete excavation, especially in proximity to the pulp. However, evidence levels are currently insufficient for definitive conclusions because of high risk of bias within studies.