In the present paper, a procedure for the determination of total arsenic and arsenic (III) in phosphate fertilizers and phosphate rocks by slurry sampling (SS) with hydride generation atomic absorption spectrometry (HG-AAS) is proposed. Arsenic (III) is determinated directly and total arsenic is determinated after reduction reaction. The procedure was optimized for the flow rate of NaBH(4), NaBH(4) and hydrochloric acid concentrations using a full two-level factorial and also a Box-Behnken design. Slurry preparation with hydrochloric acid in an ultrasonic bath allowed the determination of arsenic (III) with limits of detection and quantification of 0.1 and 0.3 microg L(-1), respectively. The precision of results, expressed as relative standard deviation (RSD), was always lower than 3%. The accuracy of this method was confirmed by analysis of certified sediment reference materials, while the procedure also allows for calibration using aqueous external standards. This method (SS/HG-AAS) was used to determine total arsenic and arsenic (III) in two phosphate rock samples and two phosphate fertilizer samples. In these samples, total arsenic concentrations varied from 5.2 to 20.0 mg kg(-1), while As (III) concentrations varied from 2.1 to 5.5 mg kg(-1), in agreement with published values. All samples were also analyzed using acid digestion/HG-AAS. Both, a paired t-test and a linear regression model demonstrated no significant difference (95% CL) between the results obtained using these two sample preparation procedures.