This study compared administration of performance tests and visual analogue scales (VAS) using a newly developed pen computer (PenC) battery with established tests using either pencil-and-paper (PP) or conventional computer. The performance of 47 subjects (23 male, age 18-45 years, weight 51-112 kg) was compared on the two systems after a dose of ethanol (0.8 g/kg up to a maximum of 60 g for males, 50 g for females) or placebo in a double-blind two-period randomized crossover study. Mean (SD) blood ethanol concentrations (breathalyser) were 94.5 mg/100 ml (21.9) at the start of the test battery (30 min post-drink) and 80.2 (13.0) at the end of the battery (75 min post-drink). Ethanol effects were found in all tests, with most outcome measures showing significant slowing or loss of accuracy. Results from the Rapid Visual Information Processing, Sentence Verification and Continuous Attention tasks show that the ethanol-placebo difference and the statistical significance of this difference are in close correspondence for the two modes of administration. The pen computer versions of these tasks may therefore be used as direct replacements for the previous versions. Digit-Symbol and maze tasks did not correspond so closely both showing differences in the speed-accuracy trade-off between the two modes. These tests, however, are sensitive to the effects of ethanol, and may be useful in their own right. Principal component analysis suggested that VAS may be grouped into two factors: (1) 'functional integrity', including measures of alertness and perceived proficiency, and (2) 'mood', including happiness and sociability. Factor 1 showed substantial effects of ethanol, while factor 2 was unchanged. There was close agreement between the results from PP and PenC for both factors as well as for the Sober-Drunk scale, which showed the expected effects of ethanol. Thus pen computer VAS perform in a similar way to the PP versions.