Nectar scents are thought to function as honest signals of reward used by pollinators, but this hypothesis has rarely been tested. Using Penstemon digitalis, we examined honest signalling of the nectar volatile (S)‐(+)‐linalool and pollinator responses to linalool in both field and laboratory settings. Because our previous work showed that linalool emission was associated with higher female fitness and that nectar is scented with linalool, we hypothesized that linalool was an honest signal of nectar reward. To assess honesty, we measured linalool–nectar associations including nectar volume, sugar amount, concentration and production rate for inflorescences and flowers in several populations. We also assessed whether Bombus impatiens, the main pollinator of P. digitalis at our sites, can use linalool as a foraging signal. We supplemented real or artificial flowers in the field and laboratory with varying linalool–nectar combinations to measure pollinator behavioural responses. We found that an inflorescence's linalool emissions could be used to predict nectar rewards in P. digitalis, but this was driven by indirect associations with display size rather than directly advertising more profitable flowers. For flowers within inflorescences there was also no evidence for an association between signal and reward. Field tests of bumblebee behaviour were inconclusive. However, in laboratory assays, bumblebees generally used variation in linalool emissions to choose more profitable flowers, demonstrating they can detect differences in linalool emitted by P. digitalis and associate them with reward profitability. These results suggest experiments that decouple display size, scent and reward are necessary to assess whether (and when) bees prefer higher linalool emissions. Bees preferred nectars with lower linalool concentrations when linalool flavoured the nectar solution, suggesting the potential for conflicting pressures on scent emission in the field. Synthesis. Our results highlight the challenges of assessing function for traits important to fitness and suggest that the perception of floral signalling honesty may depend on whether pollinators use inflorescences or flowers within inflorescences when making foraging decisions. We conclude that future research on honest signalling in flowering plants, as well as its connection to phenotypic selection, should explicitly define honesty, in theoretical and experimental contexts. Signal honesty in flowers is complex: we found that scented‐nectar is not always an honest signal in flowers of Penstemon digitalis. While inflorescences that had stronger linalool emission also produced more nectar with more sugar, we found that linalool is unlikely to predict which flowers within those inflorescences have the most nectar. The main pollinators, Bombus impatiens bumble bees, can use an association between linalool and nectar rewards to make foraging decisions, provided that the rewards are sufficiently motivating to support discriminating differences. Our results suggest that simple correlations between floral signals and rewards are unlikely to be common in nectar‐rewarding plants.