Purpose : It is undeniable that the NZ economy is heavily dependent on agriculture, especially dairy sector, but the increasing nutrient pollution discharged from dairy farms is a threat to the water quality of lakes, stream and rivers. Therefore, the dairy industry is under increasing pressure to make a commitment to improving the environmental performance of farming practices to protect water quality in waterways. Hence, farmer’s choice should be considered as one of the most important determinants of the success of policy aimed at water quality protection. It is therefore the aim of this paper to explore determinants of dairy farmers’ willingness to adopt Best manamgment practices (BMPs) for water quality protection. In addition, except for testing the commonly used determinants, such as farm characteristics, it will test for the hypothesis that spatial effects influence farmers’ choices. Design/methodology: Bayesian spatial Durbin (SDM) probit models are applied to survey data collected from dairy farmers in the Waikato Region of New Zealand. Specifically, this paper will verify the above hypothesis from two aspects. Firstly, spatial effects will be modelled according to the distance from farm to the nearest water bodies. It is assumed that dairy farmers whose farms are located close to water bodies are more inclined to be willing to adopt BMPs. Secondly, spatial effects will be presented as the existence of spatial interdependency in dairy farmers’ decision-making. It is hypothesised that dairy farmers observe or learn from nearby farmers thereby reducing the uncertainty of the performance of BMPs since BMPs are information-intensive farming techniques. Data were obtained from survey of 171 farms in the Waikato region of New Zealand; socioeconomic data were drawn from the 2013 Census. The advantage of the SDM probit model is that it allows for the inclusion of both the spatially lagged dependent variable and spatially lagged independent variables, which takes account the impact of neighbouring farmers’ decisions as well as of neighbouring farmers’ characteristics. Therefore, different from non-spatial probit models, the SDM probit model accounts for both direct and indirect effects. Significantly, the indirect effects (spatial spillovers) help to measure to what extent a change in the neighbouring farmers’ characteristics affect the adoption probability of a dairy farmer.Findings: Results show that farmers located in close proximity to each other exhibit similar choice behaviour indicating that access to industry information is an influential determinants of dairy farmers’ adoption of BMPs. In addition, these findings address the importance of farmer interactions in adoption decisions as participation in dairy-related activities are identified as an extension of information acquisition. Financial problems are considered a significant barrier to adopting BMPs. Overall, the study highlights the importance of accounting for interdependence in farmers’ decisions, which emerges as important in the formulation of agricultural-enviornmental policy. Originality/value: This paper is the first empirical study to examine the determinants of farmer’s adoption of BMPs in NZ by using spatial econometrics methods, which considers various determinants, including drivers and barriers for farmers to adopt the practices, farm and household characteristics as well as spatial issues. The results contribute to assist policy makers to specify water protection strategies. An understanding of dairy farmers’ drivers and barriers to adopting BMPs could assist policy makers to deliver support to solve the problems that are badly in the need of help. Besides, the importance of information availability in the neighbourhood network and social activities for the farmer's decision-making suggests that extension activities that address the whole community may be more efficient than targeting individual farmers to induce behavioural changes in adoption of BMPs.