The current risk assessment framework for insecticides suffers from certain shortcomings in adequately addressing the effects of low doses on off-target species. To remedy this gap, a combination of behavioural assays and in vitro cellular approaches are required to refine the precision of toxicity assessment. The domestic honey bee has long been standing as an emblematic pollinator in ecotoxicology, and once more, it provides us with a practical testing model for this purpose. First, newly emerged bees (D1) were found more vulnerable than 6 days-old bees (D6) to deltamethrin, a widely used α-cyano-3-phenoxybenzyle pyrethroid. In D1 bees, the range of doses inducing mortality was shifted towards lower values (∼2-fold) with a correspondingly lower LD50 (11 ng/bee). Moreover, at low doses that do not induce mortality in laboratory conditions, the locomotor behaviour of D1 bees was more impacted than in D6 bees. This was evidenced by an increase in immobility time and a decrease in locomotor performance across all tested doses for D1 bees (0.75, 1.5 and 3 ng/bee) during automated 21 h-long observations. Behavioural disorders are linked to deltamethrin's disruption of voltage-gated sodium channels (NaVs) functions, as quantified in cultured neuronal cells. In the presence of deltamethrin, patch-clamp experiments revealed a concentration- and a use-dependent slowing of NaV kinetics. Channel's deactivation is slowed by three orders of magnitude at 10 μM deltamethrin. Two additional phenoxybenzyle pyrethroids, including the commonly used cypermethrin, elicited quantitatively similar effects on NaV kinetics. The integration of in vitro cellular assays and behavioural assays may facilitate a deeper understanding and prediction of insecticides toxicity.