Although a potential link between silicone gel breast implants and autoimmune connective tissue disease has been suggested, none has been proven. The potential role of silicone as an immune adjuvant remains very controversial. Currently available techniques do not easily allow precise measurements of silicone in tissues. However, all compounds containing silicon (which would include silicone) can be measured accurately. The present study was designed to measure silicon levels in the fibrous capsules of patients with silicone-gel breast implants, saline breast implants and silicone inflatable penile prostheses. Baseline control silicon levels were obtained from the breast tissue of patients undergoing breast reduction, who had no exposure to breast implants. All silicon measurements were carried out using atomic absorption spectrometry with a graphite furnace. The mean silicon levels in 16 breast tissue control samples from 8 patients undergoing breast reduction varied from 0.046 to 0.742 micrograms/g dry weight, with the median mean being 0.0927. The median silicon level in capsules from 6 patients with saline implants was 7.7 micrograms/g (range 36.6). The median silicon level in capsules from 5 patients with silicone inflatable penile prostheses was 19.5 micrograms/g (range 34.8). Although the levels of silicon in capsules of patients with saline breast prostheses and penile implants were higher than in control samples, they were much lower than those from the capsules of the 58 gel implants (median 9979 micrograms/g). Of the 58 silicone gel breast implants (from 20 patients with bilateral implant removal and 18 patients with unilateral removal) which had been inserted from 1974 to 1990, 28 were intact, 8 had pinhole leaks, and 22 were ruptured. Median capsule silicon levels and ranges for all 58 implants, for intact only, for leaking, and for ruptured were: 9979 (152,000), 10,477 (88,703), 6592 (65,396), and 9922 (152,387) micrograms/g respectively. There were no significant differences in silicon levels associated with implant status, duration in situ, or year of implantation. Capsule contracture was not associated with higher levels of capsule silicon. Capsule silicon levels were about 10(6) times higher than previously assayed blood silicon levels. This may be because silicone released from implants remains localized in capsular tissue, or because blood-borne silicone is quickly excreted. Using 29Si nuclear magnetic resonance spectroscopy, no detectable silicone was found in the blood of 7 control women and 7 women with silicone-gel implants (5 with known implant rupture).