OBJECTIVES: The purpose of this study was to investigate smoking prevalence of Tri-Service recruits, and changes in smoking behaviour at 3-year follow-up, by trade group and gender. Associations with educational attainment and deprivation were also assessed. METHODS: Analysis of a survey into the health behaviours of 10 531 recruits in 1998/1999. A follow-up 3 years later measured changes in behaviour. Correlation and multiple regression was used to investigate the relationship between smoking prevalence in each trade group and both educational attainment and deprivation, using Index of Multiple Deprivation 2004 (IMD 2004) scores. RESULTS: Army recruits exhibited a significantly higher smoking prevalence (45%) than Royal Navy recruits (34%) and Royal Air Force (RAF) recruits (31%). There were marked differences between smoking levels amongst officer cadets (12%, 20% and 10% in the Navy, Army and RAF, respectively) and other rank trade groups (24-56%), with the exception of the Marines (13%). At follow up, smoking had generally increased, and in some parts of the infantry had risen to 66%. There was a clear correlation between smoking at enlistment and both educational attainment (correlation coefficient=0.7, p<0.005) and deprivation score (correlation coefficient=0.8, p<0.005). CONCLUSIONS: There were clear differences between Services, rank and trade groups in smoking prevalence at recruitment. Smoking levels increased in the 3 years after recruitment to the Armed Forces. Deprivation was more important than educational attainment in determining the smoking status of recruits.