Binomial and multinomial logit models with random effects are applied on a twenty year unbalanced panel of Norwegian households. Two binomial choices are considered: whether to smoke or not, and given smoking is chosen, whether to smoke hand rolled or manufactured cigarettes. Multinomial models are used to study the choice among four alternatives: no tobacco; manufactured cigarettes only; hand rolled cigarettes only; both manufactured and hand rolled cigarettes. Results obtained by Maximum Likelihood, and bootstrap estimation of standard errors, indicate that an increased price on manufactured cigarettes could lead consumers to switch to hand rolled cigarettes, rather than quit smoking.