A method has been developed for the determination of arsenic (As) in pharmaceutical and scalp hair samples of kidney failure patients by cloud point extraction (CPE). The scalp hair samples were subjected to microwave-assisted digestion in nitric acid-hydrogen peroxide (2 + 1, v/v). Then, dialysate and digested scalp hair solutions were preconcentrated by CPE using pyronine B as a complexing agent. The resulting complex was entrapped in nonionic surfactant (Triton X-114) prior to its determination by electrothermal atomic absorption spectroscopy. The validity of the CPE method for As was checked by analysis of a certified reference material of human hair and a standard addition method for dialysate solution. The chemical variables affecting the analytical performance of the CPE methods were studied and optimized. After optimization of the complexation and extraction conditions, a preconcentration factor of 52 was obtained for As in 10 mL of dialysate solution and acid digested samples of scalp hair. Under the optimum experimental conditions, the LOD and LOQ of As for the preconcentration of 10 mL of solution were 0.022 and 0.073 microg/L, respectively. The level of As in scalp hair samples of kidney failure patients was higher than in healthy controls.