Until recently, a long-standing assumption in the field of child language acquisition research was that parents do not correct the grammatical errors of their children. While consensus now exists that potentially corrective responses are often supplied, controversy persists as to whether the child can identify and exploit such information in practice. To address these issues, this study adopts the contrast theory of negative input as a framework for analysis (Saxton 1995). In this theory, two distinct kinds of corrective input are identified, termed negative evidence and negative feedback, respectively. The corrective potential of each category was investigated by examining the immediate effects of each on the grammaticality of child speech. A longitudinal corpus of naturalistic data (49 hours) from a single child were analysed with respect to 11 grammatical categories. The effects of negative input were compared with two non-corrective sources of input, namely positive input and adult move-ons. It was found that grammatical forms were more frequent in child speech following negative evidence and negative feedback than either of the two non-corrective sources of input. In light of these, and related, findings, it is argued that corrective input may well prove important in explanations for how the child eventually retreats from error to attain a mature system of grammar.