In a recent study of fall prevention in 301 community-living older persons (the Yale FICSIT Trial, 1990-1993), participants in the multifactorial targeted intervention (TI) group experienced significantly fewer falls than participants in the social visit (SV) control group. In the present paper, the authors explore the relation between changes in the studied risk factors and the occurrence of falling. In comparison with SV participants, TI participants showed significantly greater improvements in postural blood pressure change (p = 0.01), step length (p = 0.004), use of > or = 4 medications (p = 0.003), and unsafe tub and toilet transfers (p = 0.05), while change in balance was of borderline significance (p = 0.08). Reduction in the occurrence of falling, in turn, was at least marginally associated with improvements in balance, postural blood pressure change, step length, lower extremity strength/range of motion, and transfers. When participants were divided into tertiles based on a composite risk factor change score, a significantly higher percentage of TI participants (42%) than SV participants (22%) were in the greatest risk factor reduction tertile. Among TI participants, there was a progressively lower fall rate per person per year in the tertiles with the least, intermediate, and greatest risk reduction (0.832, 0.624, and 0.260), respectively. A similar but weaker relation between risk factor reduction and fall rate was seen in the SV group. When compared within tertiles, essentially adjusting for the amount of risk factor reduction, the fall rates among TI and SV participants in the greatest risk factor reduction tertile were identical (0.260 falls per person per year), and the rates in the least reduction tertile were similar (0.832 vs. 1.040 falls per person per year); this suggests that risk factor reduction at least partially mediated the treatment effect. These results support the feasibility of implementing and analyzing the effectiveness of a multiple risk factor reduction strategy in the aged.