BACKGROUND: Despite concerns that facilitating informed choice would decrease diabetes screening uptake, 'informed choice' invitations that increased knowledge did not affect attendance (the DICISION trial). We explored possible reasons using data from an experimental analogue study undertaken to develop the invitations. We tested a model of the impact on knowledge, attitude and intentions of a diabetes screening invitation designed to facilitate informed choices. METHODS: 417 men and women aged 40-69 recruited from town centres in the UK were randomised to receive either an invitation for diabetes screening designed to facilitate informed choice or a standard type of invitation. Knowledge of the invitation, attitude towards diabetes screening, and intention to attend for diabetes screening were assessed two weeks later. RESULTS: Attitude was a strong predictor of screening intentions (β = .64, p = .001). Knowledge added to the model but was a weak predictor of intentions (β = .13, p = .005). However, invitation type did not predict attitudes towards screening but did predict knowledge (β = -.45, p = .001), which mediated a small effect of invitation type on intention (indirect β = -.06, p = .017). CONCLUSIONS: These findings may explain why information about the benefits and harms of screening did not reduce diabetes screening attendance in the DICISION trial.