Error field induced modes have been a significant concern for next step devices. Experiments on JET, using the lower saddle coils to simulate such fields, have determined the critical scaling dependences of error field sensitivity in terms of global plasma parameters, under the conditions most relevant to next step operation. These experiments indicate relatively weak toroidal field scaling for the error field strength required to induce a locked mode, while confirming the linear density scaling observed on other devices. Correction of intrinsic errors improves low density access and assists end-of-pulse disruption avoidance. Additional heating studies have shown a strong effect of q = 2 rotation on the threshold, but a much weaker dependence on β and the H mode transition. Error field modes have also led to neoclassical tearing modes at appropriate collisionality and β. Other parameter dependences have been examined: plasma inductance has a strong effect, but plasma configuration (limiter or divertor) is less important. Application of a dimensional scaling constraint enables determination of machine size scaling, which is found to be weak. This leads to the relatively tolerable error field sensitivity expected for devices such as ITER FDR. However, there remain some important questions to be answered about the effects of rotation on error field sensitivity, and for ITER FDR some means of correction remains prudent.