This study examines predictors of locating participants that were last contacted 20 years ago using public web-search directories, in order to facilitate longitudinal environmental health research. Participants (n = 3,202) resided in four distinct geographical neighborhoods in Hamilton, Ontario during childhood; they were between 15 and 17 years old when they were last contacted in 1986. Data used for tracing included available addresses, telephone numbers, given names, and parental names. Reverse and forward search strategies were used to retrieve updated contact details. 43% of the sample was traced using online directories. Following ethical approval, participants were contacted using traced data and 29% of the original cohort was located. Predictors of locating participants were: availability of paternal names, being traced to original addresses or telephone numbers, gender (male), relatively higher socioeconomic status in childhood, and not being exposed to smoking in childhood. Where participants resided in childhood was not a significant predictor of locating participants. Although 13% of the sample was traced using forward search by name, only 4% were located. For participants traced to available addresses or telephone numbers, the difference between the proportions of traced and located participants was <3%. Prospective studies on children may benefit from including the listed names that pertain to each child's telephone number and full parental names at recruitment, thereby increasing the likelihood of locating participants using Internet resources. Integrating the use of Internet-based public directories for cohort reconstruction can reduce financial costs related to follow-up for longitudinal research. © 2009 Springer Science+Business Media B.V.