Smallholder farmers are increasingly exposed to climate variability that threatens agricultural productivity and household livelihoods. Sustainable agricultural practices (SAPs) are widely recognized as a key pathway for sustaining yields, conserving resources, and enhancing resilience. Identifying the factors that drive adoption is therefore essential for designing effective adaptation strategies. While much of the literature has focused on socio-economic determinants of technology adoption, the role of behavioral factors remains less examined. Using survey data from the Kenya Central Highlands and a Poisson regression framework, we find a non-linear, U-shaped relationship between climate concerns and adoption, with higher adoption at low and high levels of concern and lower adoption at moderate levels. The results show that subjective climate risk perceptions exert a stronger influence on adoption decisions than beliefs about climate exposure. This suggests that the way farmers interpret and internalize climate risks matter more for their behavior than whether they believe they have personally experienced climate events, underscoring the importance of addressing risk awareness and interpretation in adaptation strategies. We also find notable heterogeneity: in female-managed plots, education strongly predict adoption, while in male-managed plots, training frequency and climate concerns play important role. Additionally, certain practices such as organic inputs and soil and water conservation are more sensitive to climate concerns, while input and resource intensive practices like irrigation, intercropping, and no-tillage appear to be less influenced by climate perceptions. The findings highlight the need to integrate behavioral dimensions, particularly climate risk perceptions, into adaptation policies and contextualize interventions by gender and farm size. We emphasize that the analysis relies on cross-sectional data and therefore unobserved behavioral traits, and social-contextual factors may be correlated with climate risk perception, as well as with adoption decisions, potentially biasing our estimates. To mitigate this, we include a comprehensive set of covariates to account for socio-economic and behavioral factors that prior literature identifies as influencing technology adoption. Nonetheless, we acknowledge that some unobserved confounders may remain, and therefore we interpret our results as associations rather than causal effects. Even so, insights into how perceptions shape adoption decisions provide valuable guidance for grassroots implementers, allowing them to design awareness campaigns, training programs, and extension strategies that align with farmers’ experiences and address the thresholds or barriers that may impede timely adoption of sustainable practices.