Child welfare agents (CWA), parents reported for child abuse or neglect (RP), and exemplary parents (EP) rated 64 vignettes (61 from J. Giovannoni and R. Becerra, 1979 plus 3 additional) depicting eight subtypes of child maltreatment. Three issues related to the social construction of child abuse were examined. First was participants' perceptions of the relationship between four constructs that appear frequently in the literature: harmful parenting practices, personal definitions of child abuse, the legal definition of abuse, and the ideal legal definition. Differences between these constructs are often neglected and potentially confounded. It was hypothesized that participants would perceive more vignettes as harmful than as abuse, more as abuse than in the legal definition, and the legal definition as different from the ideal legal definition. This pattern was confirmed for CWA but only partially for EP and RP. The most significant exception was that RP believed as many vignettes were covered by the legal definition as they personally defined as abuse. Possible misinterpretation of past research is discussed in light of these findings. Second, the hypothesis that the social construction of the definition of child abuse is occurring through a top-down model of information dissemination, and therefore that CWA would include more vignettes in each of the above constructs than would EP, and that EP would include more than would RP was examined. The opposite was found. The RP defined more as abuse (personal, legal, and ideal legal definitions) than did the CWA. Findings point to the differential effect that the report and investigation process may have on the beliefs of RP and CWA. Third, the study examined differences in the degree of harm that RP, EP, and CWA assigned each of eight subtypes of abuse (sexual, physical, emotional abuse, parental drug or alcohol abuse, fostering delinquency, educational neglect, inadequate supervision, and general failure to provide). Contrary to predictions, RP assigned the highest harmfulness ratings and CWA assigned the lowest to all subtypes except physical abuse. Findings are consistent with previous research that suggests professional involvement in severe cases of child abuse may result in professionals reserving high harmfulness ratings for only extreme cases of maltreatment.