Cox model and accelerated failure time models are widely used in modelling of survival data for various diseases. This research compares the performance of Cox proportional hazards models and accelerated failure time (AFT) models using TB/HIV co-infected survival data. The tools used are AFT model plot, the log-likelihood test, Akaike Information Criterion (AIC), Log rank test for comparing all survival variables. The research established that AFT model provides a better description of the dataset as compared with Cox PH models because it allows prediction of Hazard function, survival functions as well as time ratio. Moreover, Cox proportional hazard model does not fit appropriately when compared with AFT model; thereby provide less appropriate description of the survival data. Hence, it is better for researchers of TB/HIV coinfection to consider AFT model even if the proportionality assumption of the Cox model is satisfied.