This article contributes to the related literature by empirically investigating the efficiency of nine energy and precious metal markets over the last decades, employing several pronounced models. We test for both the short- and the long-run efficiency using, in addition to linear cointegration models, nonlinear cointegration and error-correction models (ECM) which allow the efficiency intensity to change per regime. Our findings can be summarized as follows: i) futures prices are found to be cointegrated with spot prices, but they do not constitute unbiased predictors of future spot prices; ii) the hypothesis of risk neutrality is rejected and there is some evidence of time-varying risk premia; iii) the short-run efficiency hypothesis is rejected, suggesting that using past futures price returns improves the modeling and forecasting of future spot prices; and iv) the nonlinear modeling suggests the presence of two distinct regimes where in the first regime the efficiency hypothesis is supported, whereas in the second it is rejected. The empirical findings have important implications for producers, hedgers, speculators and policymakers.