Stream metabolism is a critical indicator of ecosystem health and connects stream ecology to global change. Hence, understanding the controls of metabolism is essential because streams integrate land use and could be net sources or sinks of carbon dioxide (and methane) to the atmosphere. Eleven aridland streams in the southwestern US (Arizona) across a hydroclimatic and size (watershed area) gradient were surveyed, and gross primary production (GPP) and ecosystem respiration (ER) were modeled and averaged seasonally over a period of 2-4 years. The seasonal averaged GPP went as low as 0.001 g O2m-2d-1 (Ramsey Creek in 1st quarter of 2017) and as high as 14.6 g O2m-2d-1 (Santa Cruz River in 2nd quarter of 2017), whereas that of ER ranged from 0.003 (Ramsey Creek in 1st quarter of 2017) to 20.3 g O2m-2d-1 (Santa Cruz River in 2nd quarter of in 2017). The coefficient of variation (CV) of these GPP estimates within site ranged from 42% (Upper Verde River) to 157% (Wet Beaver Creek), with an average CV of GPP 91%, whereas the CV of ER ranged from 32% (Upper Verde River) to 247% (Ramsey Creek), with an average CV of ER 85%. Among 4 main categories of hypothetical predictors (hydrology, nutrient concentration, local environment, and size) on CV and point measurement of stream metabolism, the following conclusion was made: hydrologic variation only predicted the ER and CV of ER but not the GPP or CV of GPP; light and its CV controlled GPP and its CV, respectively, whereas temperature was one of the controlling factors for ER; CV of nutrient concentration was one of the drivers of CV of GPP, nitrate concentration was correlated with point measurement of GPP and ER while soluble reactive phosphorus (SRP) concentration was only relevant to GPP; watershed area was correlated with CV of GPP, while depth mattered to both GPP and ER. My work will enhance our understanding of streams at multiple temporal and spatial scales and ultimately will benefit river management practice.