Four Diesel engines, all running on the Diesel / natural gas dual fuel mode, were dynamometer tested. Engines were selected based on their distinct construction, operating parameters and engine field use. The Diesel / natural gas substitution was varied in order to determine the dual fuel operation limits. Measurements related to particulate emissions (smoke and opacity) were also made. Correlations for indicated thermal efficiency, volumetric efficiency and engine friction were proposed. It`s suggested that those can be used on the performance estimate of engines when little or no previous test data is available. The dual fuel test results indicate that, according to engine load, part of the gas does not burn. In engines running on low load as much as 20 - 30 % of the gas supplied simply reaches exhaust manifold unreacted. A simple model for the dual fuel combustion was proposed. An empirical correlation for a combustion parameter, that arouse in the modeling was, base on test results, proposed. It`s suggested that the semi-empirical dual fuel model can be used for the performance estimate of Diesel engines, when burning natural gas, not previously tested on the dual fuel mode. Results indicate that, for engines running on air / gas ratios above 30, combustion appear to remain limited to Diesel droplets neighborhood. For richer mixtures gas combustion appear to occur on flame fronts. Audible knock was noticed on engines running on air / gas ratios below, approximately, 30. The empirical correlations proposed were used for the dual fuel conversion of three different genset engines (212, 535 and 1.570 bhp). Gensets data, when appropriate, is also reported in the present work.