This paper describes two series of load tests on slender composite columns: eight tests on concrete filled rectangular hollow steel section columns and seven tests on concrete encased rolled steel joist columns, both with end eccentricities producing moments other than single curvature bending. In all tests, normal strength concrete of grade C25/30 and low strength steel of grade S275 were specified. The main objectives of these tests were twofold: first, to provide test information for the calibration of Eurocode 4 against BS 5400 for slender composite columns and secondly to assess the accuracy of a new design method for composite columns, based on the well established steel column approach in BS 5950. Comparisons are made between predicted column strengths using the three different methods and between various predictions and test results. It is found from these comparisons that all three methods are conservative and have similar overall accuracy. For concrete filled composite columns, the Eurocode 4 method appears to be more accurate than the BS 5400 method. The proposed design method, which is based on the steel column approach in BS 5950, gives predicted column strengths comparable to those from the Eurocode 4 method. It also gives a clear understanding of the behaviour of a composite column. However, the main advantages of the proposed method are its simplicity and the savings in design time.