The purpose of this study was to evaluate a new method of in-vitro fertilization (IVF) in patients with severe sperm defects. Unlike the conventional swim-up method, spermatozoa and oocytes are placed in opposite corners of the bottom of the incubation dish so that sperm swimming is horizontal instead of vertical. Another difference between the swim-across and swim-up techniques is that the incubation medium is supplemented with 20% follicular fluid. After a randomized series (protocol I) of 15 IVF attempts had demonstrated that swim-across was more effective than swim-up in terms of fertilization and cleavage, we began a second series (protocol II) using only swim-across. A total of 124 couples with motile sperm counts less than 1 x 10(6) spermatozoa/ml of semen were included in protocol II. Clinical parameters (age, tubal damage) and number of recovered oocytes were recorded and compared in patients who did (group A: n = 94) and did not (group B: n = 74) achieve fertilization. In group A the fertilization rate was 36.7% and, out of the 94 transfers that were made, there were 21 clinical pregnancies and 12 full-term pregnancies with 16 live births. The number of oocytes collected (12 versus 7.7, P less than 0.001) and the incidence of tubal damage (50% versus 24.3%, P less than 0.001) was significantly higher in group A than in group B. Using logistic regression analysis, we showed a significant correlation between fertilization and progressive motility, percentage normal spermatozoa, number of recovered oocytes and tubal damage.(ABSTRACT TRUNCATED AT 250 WORDS)