In 2007, the New Zealand government announced a change in the legislation on tobacco on-pack warning labels, requiring text-based warning labels to be replaced by graphic health warnings (GHWs). According to this new regulation, GHWs must cover 30% of the front and 90% of the back of all cigarette and tobacco packets that are manufactured for sale in New Zealand. These new requirements took effect from 28 February 2008 with a six-month phase-in period. The aims of this thesis were to assess the effects these new GHWs had on smokers, and explore issues relating to the implementation of the new health warnings. The thesis describes a multifaceted investigation in which four independent studies were undertaken. Study One monitored the roll-out of GHWs in retail outlets before, during, and after the phase-in period. Study Two reported on a secondary analysis of a large-sample cohort study, in which smokers self-reported their knowledge of smoking-induced harms and responsiveness towards the old text-based warnings and the new GHWs. The third study collected observational data on smokers' avoidance behaviours towards on-pack warnings in public places (pre- and post-GHWs). Study Four involved a series of in-depth interviews with shopkeepers who sold tobacco, and reported on their experience in distributing the new GHWs, their support for this policy, and their observations of customers' responses towards GHWs.The data collected from Study One show that manufacturers were able to comply with the roll-out timetable for introduction of the new GHWs. Further, participants in the retailers' interviews (Study Four) did not report any problems or direct financial cost associated with implementing the new warnings. Study Two provides strong evidence that smokers were more responsive to the new GHWs when compared with the old text-based warnings. Statistically significant differences were found in all variables assessed in that study with increases observed in knowledge of specific health problems featured on GHWs, warning salience, cognitive processing of the warnings, forgoing behaviours and warning avoidance. Importantly, certain groups of smokers were more responsive to GHWs: Māori and Pacific people and those who were more financially deprived. Differences also existed according to the type of tobacco used, with those who smoked roll-your-own cigarettes being less responsiveness to GHWs than those who smoked factory-made cigarettes exclusively. This finding may suggest that GHWs printed on loose tobacco packs were less effective than those printed on factory-made cigarette packets. Finally, findings from Studies Three and Four provide further evidence that GHWs elicited stronger avoidance behaviours towards warning labels, when compared with text-based warnings. Smokers in Study Three showed higher rates of avoidance behaviours when GHWs had been implemented for a year (compared to before GHWs were implemented), and retailers in Study Four reported smokers disliked packets with GHWs and during the roll-out period requested packets with the old warnings.Consistent with the majority of the literature, the research in this thesis show that GHWs created greater impacts on smokers than text-based warnings. The findings also suggest that GHWs have the potential to contribute to reducing smoking-induced inequality by eliciting greater responses among high smoking prevalence disadvantaged groups. Findings from this thesis have important policy implications that will help future development of warning label requirements in New Zealand, as well as other countries that are considering introducing GHWs.