Benign prostatic hyperplasia (BPHs) affects more than half of men who are at least 60 years old. Transurethral resection of the prostate (TURP) is the most common treatment. This article estimates the average costs of providing a TURP to a patient that chooses the procedure and the expected cost of a TURP irrespective of the actual treatment decision. It utilizes a modified two-part model. It first estimates the likelihood of receiving the TURP given a BPH diagnosis using a logit regression. The second step is to estimate the (log of) TURP cost for the population that actually received the procedure. For the TURP decision, data were extracted from the National Health Interview Survey (NHIS) 1986–1994 on 502 men with a diagnosis of BPH. For the cost model, HCIA discharge abstract data were used from five states on 26640 men who actually received a TURP in 1994. TURP was identified via CPT codes in the NHIS data. The cost of TURP was constructed using patient hospital charges and hospital-specific cost to charge ratios. It was found that patient characteristics and comorbidities are important determinants of the decision to receive, and costs of, TURP. Pharmacological alternatives to TURP significantly affect the likelihood that a person will choose the procedure. The predicted unconditional cost of TURP ranged from US$8908 to US$3832 (by state); the predicted conditional cost of TURP ranged from US$1163 to US$750. The results suggest that the cost associated with providing a TURP for the average man with BPH would be US$492 to US$1163, depending on the state of residence. It was found that estimating the costs using simple regression or ANOVA techniques will lead to biased results due to sample selection (which in this case range from US$3832 to US$8908).