Time series data were collected twice daily for 62 days from 10 individuals on three variables related to smoking habit strength: number of cigarettes smoked, salivary cotinine, and carbon monoxide. The two purposes of this study were: (a) to evaluate which time series model(s) best fits each of the measures; and (b) to determine which model of nicotine regulation is consistent with the data. Three models of nicotine regulation were considered: (a) nicotine fixed effect; (b) nicotine regulation; and (c) multiple regulation. These models provide different predictions about the size and direction of the lag-one autocorrelation. Each measure was described in terms of one of a family of time series models represented mathematically as ARIMA (p, d, q). Models varied by individual, but a single model described the majority of subjects for all three variables. The clearest model identification was for the number of cigarettes smoked where an ARIMA (1, 0, 0) model with a moderate to strong negative dependency fit the majority of the subjects. This provided strong support for the multiple regulation model. An appendix provides a brief review of time series methodology.