Vicarious experience gained through seeing women breastfeed may influence infant feeding decisions and self-efficacy. Our aim was to measure the attributes of seeing breastfeeding and to investigate how these relate to feeding intention (primary outcome) and behaviour (secondary outcome). First, we developed a Seeing Breastfeeding Scale (SBS), which consisted of five attitudes (Cronbach's alpha of 0.86) to most recently observed breastfeeding: ‘I felt embarrassed’; ‘I felt uncomfortable’; ‘I did not know where to look’; and ‘It was lovely’ and ‘It didn't bother me’. Test–retest reliability showed agreement (with one exception, kappas ranged from 0.36 to 0.71). Second, we conducted a longitudinal survey of 418 consecutive pregnant women in rural Scotland. We selected the 259 women who had never breastfed before for further analysis. Following multiple adjustments, women who agreed that ‘It was lovely to see her breastfeed’ were more than six times more likely to intend to breastfeed compared with women who disagreed with the statement [odds ratio (OR) 6.72, 95% confidence interval (CI) 2.85–15.82]. Women who completed their full-time education aged 17 (OR 3.09, 95% CI 1.41–6.77) or aged 19 (OR 7.41 95% CI 2.51–21.94) were more likely to initiate breastfeeding. Women who reported seeing breastfeeding within the preceding 12 months were significantly more likely to agree with the statement ‘It was lovely to see her breastfeed’ (P = 0.02). Positive attitudes to recently seen breastfeeding are more important determinants of feeding intention than age of first seeing breastfeeding, the relationship to the person seen and seeing breastfeeding in the media.