MEDLINE, EMBASE, The Cochrane Oral Health Group's Trials Register, The Cochrane Central Register of Controlled Trials (CENTRAL), UK National Research Register, Australian New Zealand Clinical Trials Registry (ANZCTR), Database of Abstracts of Reviews of Effectiveness (DARE), ISI Proceedings for relevant conference abstracts. The search strategy used keywords but not subject heading terms. A number of relevant journals were hand searched (seven most recent years) and authors were contacted in the absence of complete data. Randomised controlled trials (RCT) or controlled clinical trials (CCT) reported in English only, that compared platform-switched to platform-matched implants were eligible. A minimum of 10 implants had to have been placed in the platform-switched group (it is unclear if there was a minimum for the comparison group) and they had to have been followed up for a minimum of 12 months. Primary outcome was marginal bone level changes. Secondary outcome was implant failure rate. Data were extracted by more than one author using a data extraction form. Quality assessment was done using the Jadad scale. Meta-analysis was conducted using fixed effects model in the absence of significant heterogeneity, and the random effects model where heterogeneity was greater. Statistical heterogeneity was assessed using the chi(2) and I(2) tests. Sensitivity and subgroup analyses were planned to identify any potential causes of heterogeneity. Ten studies including 1239 implants were included and all were published 2007-2010. Seven were RCTs, three were CCTs. Range of observation was 12-60 months. Methodological quality was assessed as 'satisfactory'. Chi(2) =126.79 (P <0.0001), I(2) = 91% indicating significant statistical heterogeneity. Thus the random effects model was used to synthesise the data. Bone loss in the platform-matched implant group was greater with a mean difference of -0.37 mm (95% CI -0.55 to -0.20, P <0.0001). This is based on the longest follow-up interval from each trial (therefore could be anywhere between 12 and 60 months). There was no significant difference in implant failure. Platform-switching may preserve vertical crestal bone levels more than platform-matching when placing implants.