In this paper we analyze the persistence of US REITs over the period 1990-2005. By employing a novel methodological approach we shed new light on whether investors can pick winners by simply looking at past performance. The private real estate industry is notorious for its lack of informational transparency. Hence, professional portfolio managers with superior skills and access to scarce information might be capable of persistently outperforming their competitors in this market. We conduct a two-stage approach to investigate this matter. First we analyze whether outperformance, quantified by a conventional Jensen alpha in a single index model specification, is persistent over time. Our first results indicate that future outperformance is indeed a function of success from previous eras. Next we calibrate a principal component model in which we identify the structural variation in performance across different real estate subsectors like the office, residential and retail market. We construct subsector indices that capture the performance of these individual submarkets and run multivariate regressions in order to correct for these sectors characteristics. When repeating our persistence analysis we find that persistence in outperformance increased when including the sector specific return effects. Apparently, some REIT-managers are more successful in selecting and managing real estate objects that continue to outperform the portfolios of competitors in their own line of industry.