To examine the relationship of socio-demographic characteristics, psychological factors, knowledge, attitude and behavior towards obesity among Metropolitan Waterworks Authority (MWWA) officers, a cross-sectional study was conducted between July and September, 2004. Two hundred and eighty-eight obese [body mass index (BMI) > or = 25 kg/m2] and 106 non-obese persons, aged 20-60 years, were recruited as study subjects. Data were collected by a self-administered questionnaire, comprised of three parts: socio-demographic; psychological factors (depression and stress); and knowledge, attitude, behavior related to obesity. Univariate analyses and Logistic regression models were used to study the association between obesity and possible risk factors. The results demonstrate significant associations between older age and obesity. Volunteers in the age groups of 40-49 and 50-59 years had a significantly higher risk of being obese than the age group of less than 40 years (adjusted OR = 3.4, 95% Cl = 1.1-11.1 and adjusted OR = 10.4, 95% CI = 3.3-32.7, respectively). Volunteers with unhealthy behaviors were at significantly higher risk than those with healthy behaviors (adjusted OR = 10.3, 95% CI = 2.0-52.4) while persons with moderately healthy behaviors also had increased risk, but to a lesser extent (adjusted OR = 4.5, 95% CI = 1.7-11.4). There were no associations between psychological factors and obesity in this group of volunteers. When we focused on whether they consumed more food when they were stressed, it was found that the obese consumed significantly more food during stress (p-value = 0.003). Watching television, videos, or playing computer continuously for more than 3 hours, were significantly associated with obesity. We conclude that although the obese have a good knowledge and attitude towards obesity, they still practise unhealthy behavior, have a sedentary lifestyle, and over eat when they are stressed. Future research regarding behavioral modification should be implemented at both community and country levels.