The majority of research into beam-column knee joints has been conducted with monotonic loading. Many of these joints failed to reach their member moment capacity, especially under opening moments, while a few cyclic knee joint tests have been completed in the United States this decade. This paper describes the cyclic testing of 8 small knee joints designed to the 1995 New Zealand Concrete Standard. In addition two joints designed and detailed to the 1965 N.Z. Concrete Code were also tested. Joints with U-bar anchorages performed better than joints with standard 90 degree hook details on beam and column bars. The current Concrete Standard (NZS3101:1995) designs usually attained their nominal moment capacity in both directions up to and including ductility 4 displacements, but subsequently strengths fell off at higher ductilities. Joints with extra diagonal bars across the inner comer were able to sustain their nominal member strengths to higher ductility levels, especially under opening moments. A maximum horizontal joint shear stress of 0.12 f’c, for knee joints, in ductile frame buildings is recommended, where this limit is 60% of the current NZS3101:1995 Standard recommendation. An approximate 25% degradation of the joint shear stress occurred as displacement ductility factors increased from 1 to 8. The 1960's designed joints behaved poorly, as expected, with joint shear and anchorage failures occurring, in both moment directions, at strength levels below the beam's nominal strength. A maximum joint shear stress of only 0.072 f’c was reached and this fell to about a third of that stress between displacement ductility factors of 1 and 4 under closing moments.