Very few population surveys of aerobic fitness have been undertaken largely due to the difficulty of carrying out physiological tests in households. Successful surveys in Canada used a graded step test (Shephard et al. 1976) and the present study aimed to develop a suitable test for UK conditions. Preliminary laboratory studies indicated that a shallow double step (step height 16.5cm) was required for many subjects of average or below average fitness. Under laboratory conditions we looked at the relation between rate of stepping and oxygen consumption ([vdot]O₂) and found it to increase linearly with power ([vdot]O₂ [l/min] = 0.364 + 0.0191*power [W]), very similar to previous studies using steps (e.g. Nagel et al. 1965). Thirty four subjects of varying age, gender and fitness then carried out the step test and also had their maximum oxygen consumption ([vdot]O₂max) determined using a graded treadmill test. During the step test the subjects stepped at an initial cadence of 65, 75 or 85 movements per minute for 3 min and heart rate (HR) was measured using a Polar heart monitor over the last 20 s. Without breaks, subjects continued at 2 increased cadences (+15 and +30 from initial) for 2 min with HR determined during the last 20 s of each. The initial cadence was set by a calculation based on the age, weight, height and gender of each subject with the aim that the final cadence would produce a HR less than 80% of predicted maximum. Heart rate increased linearly with cadence, i.e. with power production, and the HR:power relationship was extrapolated to estimated maximum HR (220-Age). [vdot]O₂max was then determined from the relationship between power and [vdot]O₂ given above and was found to correlate well with [vdot]O₂max from the treadmill test (r=0.80; average difference +0.35 l/min; 95% limits of agreement +1.47, -0.77 l/min). Research nurses from the Health Survey for England took the step (weight 6 kg) to 29 homes of widely varying type (e.g. top floor flat, terrace houses, isolated farmhouse). They used their survey laptop to run the test, including producing the timing pulses, and for manual data entry. Each visit involved a single nurse running the entire test. The nurses found the test interesting to carry out and the reactions of potential subjects were positive. The test took 20 min, but with extra time for setting up, dismantling and giving explanations it is estimated that under normal survey conditions the test would take 30 min. The equipment performed reliably under field conditions. [vdot]O₂max values agreed on average with those estimated in setting the initial step rate (based on age, gender and height), the average difference being 0.2 l/min (limits of agreement +1.3 and 1.1 l/min). However, the present step needs to be made lighter and simpler to handle before it could be used in a large scale survey. In conclusion, a simple step test can be used for measuring fitness during a household survey. However, around half an hour is needed per test and the step itself needs to be easily portable and straightforward to assemble and pack away.