The relation between pre-treatment blood-pressure and the fall in pressure after treatment was examined for most classes of antihypertensive drugs. Positive correlations were demonstrated for all drugs, for placebo, and for bed rest. This suggests that for all manoeuvres response is related to the height of the pretreatment pressure. Substitution of the pre-treatment and achieved pressures by random numbers reveals that positive correlations are mathematically inevitable and do not indicate any action on a basic mechanism of essential hypertension. After statistical correction for mathematical associations between the variables the apparent effects were generally lost. A correlation between the pre-treatment value of any variable and its change after a therapeutic intervention thus may not be valid.