Within the context of the bradyphrenia debate, two experiments designed to measure the cognitive speed of Parkinson patients (PD) were conducted with 58 subjects. The experiments took the form of a high-speed memory scanning task using memory sets consisting of one-six words or one-six abstract figures. In a visual discrimination task, two simultaneously presented abstract images had to be compared, the complexity of the images being varied through four stages. Motor response was constant, reaction time was the dependent variable. PD differed from matched controls in the level (significant only in scanning) but in neither experiment in the slope of reaction time curves. Interaction group x complexity (MANOVA) was not significant. Correlations between disease parameters and a 'complexity effect' measure were low throughout. However, the link to general intelligence was significant. The findings contradict the conventional bradyphrenia hypothesis.