Fifteen beams with-unbonded tendons consisting of I and rectangular sections with different amounts of prestressed and non-prestressed reinforcement, were tested under short-term and sustained loading and a combination of sustained loading with intermittent short-term cyclic loading(combined loading). Two additional ordinary reinforced concrete beams were tested under combined loading for comparison purposes. Results indicated that a noticeable amount of non-recoverable residual deflection occurred due to the effect of cyclic load. The cause was believed to be non-recoverable creep strain and increased creep rate under cyclic loading. An analytical method was formulated for calculating the short-term deflection of unbonded partially prestressed beams. The deflection was calculated by integration of curvature based on the recommendations of CP110, Appendix A, with certain modification. The computed results agreed well with the experiments. The experimental deflection was also checked against the computed results according to the Model Code and the ACI Code. The former was found to be unconservative for unbonded I-section beams. The ACI Code I-effective formula might require modification of the power in order to produce consistently conservative results. Moreover, the ACI simplified formula for calculating the long-term deflection was unconservative for unbonded beams both for sustained and combined loading. The flexural strength of the test beams was greater than predicted by the CP110, Tam-Pannell and the ACI Code methods mainly due to underestimation of the tendon stress at ultimate moment. The stress in the tendon reached the 0.2% proof stress and the stress in the non-prestressed steel sometimes reached the 2.5% proof stress. The friction between the tendon and the concrete caused localised stress change and hence increased the strength of the unbonded beam significantly.