This paper is about to estimate the health insurance demand function in the Iranian urban area. Considering to the censored data used in this study, in the outset, a simple Tobit model has been used to avoid the censoring bias. One of the main weaknesses of the Tobit analysis appears in presence of heteroskedasticity. To measure this weakness’ significance, the demand model has also been estimated using quantile regressions, namely Powell’s estimators, and coefficients at different quantiles has been compared with those form the simple Tobit estimate. Besides, the comparisons between coefficients confidence intervals generated using the Boot-strap method for relevant quantils and simple Tobit model’s confidence intervals do not imply a serious heteroskedasticity problem. According to estimation results, the income elasticity of private health insurance demand is positive, characterizing it as a normal service. Controlling for other factors, consumers evaluate private health insurance a substitution for its public counterpart and household’s demand for this service is a concave function of household head’s age so that the demand is maximized at 55 years old. A similar story applies to the relationship between the demand for private health insurance and household head’s education level such that earning a master’s degree maximizes the demand. The wealth level, measured by some proxies in this study, carries a significant positive effect on the demand. Estimations also imply that households set their private health insurance demand based on expected medical costs. One of the interesting results, which can derive other researches in this area, is the fact that households with retired heads significantly demand more private health insurance than others.