Identifying twins for a population-based register can be achieved through birth records or community surveys. We studied the feasibility and effectiveness of different methods of identifying and recruiting twins to establish a population based register. To trace twins a population survey was carried out using an interviewer administered questionnaire. We also inspected the birth registration certificates at a divisional secretariat reported from a specified hospital between the years of 1985–1997 and compared it to the birth register of this same hospital. To recruit twins a random sample of 75 twin pairs (150 twins) identified at the Divisional Secretariat were contacted through the post and 25 twin pairs (50 twins) were personally visited. The prevalence of twins was 6.5 twins per 1000 people in the area surveyed. The twinning rate at the hospital was 18.92 twins per 1000 births. A discrepancy of 38 multiples births between the hospital labour room records and those registered at the DS was noted. The response from the postal invitation for recruitment was 59% and the response from the personal invitation was 68%. (Difference 9.4% 95% CI; 7.06–11.73). Community survey and systematic inspection of birth records either at the hospital or the birth registration office was an effective method to trace twins. Once traced, personal contact was more effective than the postal invitation for recruitment of younger twins. A cost-effective approach would be to use a postal coverage followed by personal contact for non-responders. The alternative method, community coverage, would have financial implications.