This study investigated the regulation of soil surface respiration (R(s)) in a grazed pasture, in New Zealand. The interactive responses of R(s) to soil temperature (T(s)) and root zone volumetric water content (theta) were quantified and modelled using soil cores in controlled conditions. Chamber measurements of R(s), made at a field site approximately twice a month throughout a growing season, were used to validate the model. A closed gas-exchange system was used in the field to measure ecosystem respiration (R(e)) and R(s) in order to partition the contribution of above- and below-ground respiration. This result was then applied to partition nighttime measurements of R(e) using eddy covariance into R(s) and above-ground respiration (R(a)). Eddy covariance soil surface respiration (R(sEC)) estimates were compared to modelled estimates of R(s). The response of R(s) to T(s) could be described using an Arrhenius-type function, while the response of R(s) to normalised theta(theta(n)) was described by linear relationship. The model showed that when theta(n) was >0.9, R(s) remained constant. In the field, R(s), was strongly influenced by T(s), which ranged from a minimum of 12.2 degrees C to a maximum of 20.1 degrees C throughout the measurement period. During the year, theta was rarely low enough to limit R(s). The base value of respiration at 10 degrees C (R(10)) at the field site was 0.21 mg CO(2) m(2) s(-1). The model was able to explain half of the variability observed in the field measurements of R(s). Total R(e) was comprised of 84% R(s) and 16% R(a). For nights with valid eddy covariance data, the estimate of R(sEC) from measurements of nighttime R(e) multiplied by the fraction of below-ground to total respiration was 48% less than the estimate from the model. Total growing season R(s) estimated using the model was 1.94 kg (C) m(2). (C) 2008 Elsevier B.V. All rights reserved.