Two ‘individual’ variables (level of imagery, locus of control) were investigated for their effects upon subjects' ability to learn and generalise downward control of blood pressure via biofeedback procedures. Also investigated was the effect of visual numerical feedback versus no-feedback combined with instructions. Subjects were volunteers with normal blood pressure who were randomly assigned to feedback of systolic blood pressure, feedback of diastolic blood pressure or no-feedback under an interrupted time-series design which consisted of an adaptation period, baseline, training, rest, and post-test (generalisation of task without feedback). Results indicated that neither of the two ‘individual’ variables was associated with any significant reductions of blood pressure within either the training or generalisation periods. Feedback was negatively associated with training effects. Findings are discussed in terms of the ease of comprehension of numerical feedback of a fluctuating variable, with suggestions for future research using fine-grained and stable measures of cardiac performance.