Three separate studies were conducted in the Grain Quality and Nutrition Laboratories at the Bangladesh Rice Research Institute (BRRI), Gazipur, Bangladesh. Two were dose–response studies of zinc added to soaking water at increasing concentrations. The other tested two different absolute amounts and ratios of iron and zinc added simultaneously to soaking water (150:100 and 200:150 mg L−1) as selected on the basis of results of previous studies. The iron:zinc study was conducted using two major Bangladeshi rice varieties representing the two main growing seasons: one boro rice variety (BRRI dhan29) and one aman variety (BR11), while the zinc dose–response studies used BRRI dhan29. The change in zinc content, or zinc and iron content, was deter-mined in raw rice, and after washing and cooking to calculateretention.Preparation of paddy samples Single batches of BRRI dhan29 and BR11 rice paddy were divided into samples of 200 g. All samples were prepared in duplicate, and were individually washed three times in distilled deionized water and drained prior to soaking.Preparation of soaking water solutions Distilled deionized water, food-grade ZnSO4.7H2O and sodium iron (III)–EDTA were used to prepare the zinc and iron:zinc solutions. For the zinc dose–response studies, soaking solutions were prepared to obtain concentrations of 0, 100, 150, 200 and 300 mg zinc L−1 in the first study, and 0, 500, 700, 900, 1100 and 1300 mg zinc L−1 in the second study. For the iron:zinc study, two separate solutions of 150:100 mg L−1 iron:zinc and 200:150 mg L−1 iron:zinc were prepared, and a 0:0 control solution. Soaking and parboiling Paddy samples in all experiments were subjected to a 2 min pre-steaming, followed by soaking and subsequent pressure par-boiling for 10 min. These conditions were chosen to represent the most usual conditions used in large-scale, commercial, automatic mills in Bangladesh. Pre-steaming was done by placing the paddy in an autoclave at 1000C, 0 g cm−2 for 2 min, with a water vessel as a source of steam, followed by soaking in the prepared solutions. The pH was measured in the soaking water solution before and after adding the paddy, and at 3 h intervals. After 9 h,excess water was drained from the vessel and the soaked paddy was steam parboiled in a pressurized autoclave (10 min, reaching1210C, 1 g cm−2). The second zinc dose–response study was also performed under conditions representing those commonly used in semi-automatic mill operations in Bangladesh. These use ambient soaking water conditions without pre-steaming and an open parboiling system. Paddy was soaked in the solutions at ambient temperature (approximately 25?C) for 24 h, after which excess water was drained and the paddy was steamed in a non-pressurized auto-clave at 1000C for 30 min. Dehusking and milling Outer husks were removed from dried paddy using a Satake testing husker (model THU-35B, Satake Corp., Hiroshima, Japan) with rubber rollers coated with PVC compound to avoid mineral contamination. The dehusked brown rice was milled using a Grain mantester mill (model 60-220-50-DT, Grain Machinery Manufacturing Corp., Miami, FL, USA). The dehusked rice was milled to ‘polished’ rice (i.e. 10% degree of milling). Prior to mineral analysis, all samples were thoroughly cleaned to remove exogenous residue from the surface of the rice grains. Retention with washing and cooking Sub-samples of rice were subjected to washing and cooking proto-cols representing methods commonly used in Bangladesh. Rice and distilled deionized water (1:3 by weight) were stirred in a beaker for 30 s, the excess water was poured off and this was repeated twice more. The washed samples were cooked following a standard excess water cooking method,7in which rice was cooked in a 1:10 ratio by weight in distilled deionized water by adding rice samples to boiling water, and cooking for 15 min.Excess cooking water was drained off and cooked rice was dried to constant weight. Zinc and iron content determination The zinc and iron contents of rice samples were determined induplicate in whole grains using X-ray fluorescence (X-Supreme8000, Oxford Instruments, Singapore) following a previously described method.8A previous comparative analysis of zinc content inn=40 raw rice samples with 5.5–23.0 mg zinc kg−1 determined by atomic absorption spectrometry (AAS; modelAA-6800, Shimadzu Corp., Tokyo, Japan)9and X-ray fluorescence resulted in a high correlation coefficient (r=0.982) and a mean±SD and maximum difference of 0.1±0.7 and 1.8 mg kg−1, respectively (KA Kabir, personal communication). All samples in this study were measured in duplicate or triplicate. A standard pooled sample of raw polished rice was included in the analysis after every ninth sample; this analysis produced a mean±SD of 26.2±0.7 mg zinc kg−1 (n=24) with a 2.6% coefficient of variation and 13.4±0.2 mg iron kg−1 with a 1.8% coefficient of variation, indicating a high degree of reproducibility. For the zinc dose–response studies, analysis of variance (ANOVA) with least significant differences post hoc test was used to differentiate means at different zinc treatment levels with 95% confidence. Estimation of impact on dietary zinc intake adequacy The potential impact of parboiled zinc-fortified rice on the adequacy of zinc intake by 240 children 24–48 months of age and 240 non-pregnant women in Bangladesh was simulated using an existing dietary intake database. The rice zinc content was modified in the food composition database using the control (baseline), 300 and 1300 mg zinc L−1 results for cooked rice after adjusting for fresh weight water content. Different coverage levels of par-boiled fortified rice were simulated by randomly selecting 15%,50% and 80% of women and children for application of the modified rice zinc content. The zinc intake from rice was recalculated and added to the zinc intake from all other food items reported.The National Cancer Institute method11wasusedtoobtaindis-tributions of usual zinc intake by removing within-person variability, and to determine the prevalence of individuals with zinc intake below the estimated average requirement (EAR) for zinc. The mean phytate: zinc molar ratio of the population diet was used to determine the overall zinc bio availability and the appropriate EAR for comparison, where a mean phytate:zinc≤18 rep-resents a moderate-bioavailability diet and>18 represents alow-bioavailability diet. The EAR for 1- to 3-year-old children is 2mg d−1 for both levels of bio-availability, whereas for women consuming a moderate- or low-bioavailable diet the EARs are 6 and 7mg d−1, respectively.