Regular weight-bearing physical activity has been widely recommended for adult women and may be beneficial in preserving bone mineral density (BMD). However, there is conflicting evidence regarding the effects of resistance training on BMD in premenopausal women. Novel systematic review and meta-analysis evidence is presented on the effects of progressive high-intensity resistance training on BMD in premenopausal women. Structured computer searches of MEDLINE, EMBASE, PubMed, Web of Science, SportDiscus and Evidence Based Medicine Reviews Multifile were undertaken along with hand-searching of key journals and reference lists to locate relevant studies published up to September 2004. Criteria for included studies were published controlled studies and randomised controlled trials (RCTs) evaluating the effects of progressive, high-intensity resistance training studies on BMD in premenopausal women. Two authors reached consensus on all included and excluded studies. Study outcomes for analysis were radiographic BMD assessment from first follow-up at lumbar spine and femoral neck. Primary outcomes for analysis were absolute changes in BMD g/cm(2) at lumbar spine and femoral neck. Relative changes (percentage change) in BMD at lumbar spine were also assessed. Data were extracted from studies including study design, participant characteristics and treatment mode, intensity and duration, using electronic data extraction forms. Where necessary, relevant information was obtained by contacting study authors. Methodological quality of studies was assessed using a well recognised three-question instrument designed to assess bias. Informal assessment for small sample study effects and potential bias was undertaken through visual inspection of funnel plots. The weighted mean difference method (inverse of the variances) was used for combining study group estimates. Quantification of the effect of heterogeneity among study outcomes was assessed using the I(2) statistic. Random effects and fixed-effect models were applied according to observed study heterogeneity. Comparisons resulting in I(2) > 50.0% were considered heterogeneous. Where heterogeneity was observed, a random effects model was applied. Pooled estimates of effect were calculated using the Cochrane Collaboration's Review Manager (RevMan) 4.2.1 software.High-intensity progressive resistance training was shown to be efficacious in increasing absolute BMD at the lumbar spine (p < 0.00001) but not the femoral neck (p = 0.78) in premenopausal women. The weighted mean difference (WMD) using a fixed-effect model for six controlled trials investigating the lumbar spine BMD change was 0.014 g/cm(2) (95% CI 0.009, 0.019; p < 0.00001). The relative BMD change for this site was 0.98% (WMD [random effects], 95% CI 0.49, 3.91%; p = 0.04). In contrast, studies evaluating femoral neck BMD changes showed no significant BMD change (WMD [fixed effect], 0.001 g/cm(2) 95% CI -0.006, 0.008; p = 0.78). Funnel plot inspection of lumbar spine effects indicated that smaller studies demonstrated larger treatment effects. An asymmetry towards studies with positive BMD outcomes was also noted. The methodological quality score of all included studies was low and no study presented a valid intention-to-treat accounting for participant drop-out (attrition). As such, the modest overall treatment effects for resistance training on BMD among premenopausal women observed in this review may be biased and should be interpreted with caution. It is concluded that further RCTs of resistance training of sufficiently long duration and providing optimum type, intensity and volume of loading, with intention-to-treat analysis are now required.