This paper demonstrates the application of survival models to CTD (Cumulative Trauma Disorder) studies. Survival analysis techniques are usually applied to the analysis of prospective epidemiological studies examining related risk factors. However, survival models have never been used in CTD etiology, perhaps due to mathematical complexities imposed on survival analysis techniques and its interpretation. The other reason might be the fact that it has not been considered as serious as other diseases usually studied in epidemiology. However, the conditions and assumptions of survival models fit completely into CTD studies (the existence of concomitant variables, a heterogeneous study population, censored observations etc.). Thus it is inappropriate to analyze CTD problems (specifically etiology and prevention) using typical statistical technique (ANOVA or ordinary regression). In this study, 143 subjects participated from an automobile carpet manufacturing plant which was experiencing a high number of CTD cases (107 cases, 1988–1989) and two groups of potential risk factors were examined. They were mainly categorized into personal and job-related information. This information was collected through survey questionnaires and video-taping. As the first step of analysis, survival, hazard and probability density functions were estimated. The estimated survival function shows that CTD incidence rate remained relatively constant, fluctuating 5–15%, through the first 12 years. Also, univariate associations between survival time and individual risk factors were tested using log rank and Wilcoxen rank test. Finally, the CTD data was fitted to ‘Proportional Hazard Model' (the most generalized survival model with distribution-free baseline hazard function). This model explains 75% (R2=0.75) of CTD data with the following covariates; cycle time per part, number of meals a day, dominant hand, general attitude, hand posture, degree of physical fitness, hobby and job title. The feasibility of the Proportional Hazard Modeling was investigated by test and plot. The test was conducted under the global null hypothesis about the significance of the overall model and t vs. log(-log(S(t))) was plotted to check with the assumption of proportionality. Both results ascertain the feasibility of Proportional Hazard Modeling for CTD studies.