Use of the nylon endocervical brush has been shown to increase the proportion of Papanicolaou smears with endocervical cells in a variety of populations. Each endocervical brush is approximately 100 times more expensive than the cotton swab that has been used routinely for cervical cytology collection. The authors sought to determine prospectively whether the endocervical brush would be cost effective by reducing the need for repeat sampling due to Papanicolaou smears lacking endocervical cells. During November and December 1990 (baseline phase), all Papanicolaou smears were performed with Ayre spatulas and saline-moistened cotton swabs. Endocervical brushes were not available. In January and February 1991 (experimental phase), endocervical brushes and Ayre spatulas were provided in each examination room with instructions that the endocervical brush be used on all nonpregnant patients. In March and April 1991 (control phase), Ayre spatulas and saline-moistened cotton swabs were used. Endocervical brushes again were made unavailable. Only those costs that differed between the control and experimental groups were evaluated. Costs that were identical in both sets of patients, such as the initial clinic visits, initial Papanicolaou smear and Ayre spatulas, were excluded from analysis. Cost of the endocervical brushes was $29.00 per 100, while the cotton swabs were $0.30 per 100. The cost of nursing time for follow-up on less-than-optimal Papanicolaou smears (identification, notification of patients and charting) was estimated to be $3.00. The cost of a short visit to the clinic for a repeat cervical smear was $23.00. The Pathology Department's charge for a Papanicolaou smear, including technical and professional charges, was $28.00. In November and December 1990, 467 cervical cytology specimens were collected from nonpregnant patients. One hundred fourteen smears (24 per cent) contained no endocervical cells. In January and February 1991, 511 cervical cytology specimens were collected. Thirty-seven (7 per cent) had no endocervical cells. In March and April 1991, 561 cervical cytology specimens were collected. One hundred fourteen (20 per cent) had no endocervical cells. The difference between the experimental and control groups was highly statistically significant (P < .001). Assuming 550 cervical smears in any 2-month period, the savings by using the endocervical brush rather than the cotton swab, including the cost of repeating smears lacking endocervical cells, showed that, over a 2-month period, routine use of the endocervical brush would save $3719.98. Over 1 year, the net savings would be $22,319.88.