The mature coconuts were collected from local markets of Dinajpur, Bangladesh. All chemicals used were of analytical grade and were purchased from Sigma Aldrich Chemical Co. (St. Louis, MO, USA). Collected coconuts were washed by running potable water and broken into two halves. A coconut scraper (ODIRIS Grater–B5, Odiris Engineering Co. Ltd, Colombo, Sri Lanka) was used to collect grated coconut. Coconut milk was extracted with and without testa. The grated coconut was steam blanched for 10 minutes and milk was extracted using hydraulic press (SKY004, Sakaya Automate Co. Ltd, Bangkok, Thailand) without adding water. Calculated amounts (coconut milk to water was 2: 1 w/w) of warm distilled water was added to the extracted coconut milk. The mixture was screened through a clean wire mesh sieve (150SIW.300, Endecotts Ltd, London, UK). The extracted coconut milk was separated into four samples, viz. sample–1: coconut milk from endosperm without testa, sample–2: coconut milk from endosperm with testa, sample–3: pasteurized coconut milk from endosperm without testa and sample–4: pasteurized coconut milk from endosperm with testa. Pasteurization of coconut milk was done in a thermostatic water bath (VS–310SWR, Vision Scientific Co. Ltd., Daejeon, Korea) at about 72 ± 1°C for 15 minutes. After that, we poured the coconut milk into low–density polyethylene (LDPE) of 28 µm thickness (Moyls et al.,1992) and sealed it by a hand sealing machine (PFS–200, Yongkang Golden Sky Imp.& Exp. Co., Ltd., Zhejiang, China). It was then stored at refrigeration temperature (4°C) and the analyses mentioned afterward were carried out on 0, 7, 14 and 21 days. Proximate composition serves as an important base to study the nutritive quality of coconut endosperm and milk. Samples were analyzed for proximate composition using standard methods of analyses of AOAC (2005). The moisture content of the samples was determined by oven (VS–4150ND, Vision Scientific Co. Ltd., Daejeon, Korea) drying method (at 105°C for 6 hours), whereas, the total ash content was determined using the muffle furnace (5300A30/F6010–TS, Thomas Scientific, Swedesboro, USA) at 550°C. Meanwhile, the solvent extraction method was used to determine the fat content by the soxhlet apparatus (GI–1706–A, Garg Process Glass India Private Ltd., Mumbai, India). The protein (N × 6.25) content of the samples was determined by the Kjeldahl (KjelMaster K–375, BUCHI, Flawil, Switzerland) method and carbohydrate was calculated by the difference method (Amon et al., 2014). The total energy value of coconut milk samples was determined as follows: Energy (Kcal/100g) = [(9 × %Fat) + (4 × %Protein) + (4 × %Carbohydrate)]. The %Total solids content was calculated by using the data obtained during moisture estimation using the following formula: %Total solids = 100 – %Moisture content. Titratable acidity (TA) was determined by titration of a known quantity of sample (10 mL) against 0.1 N sodium hydroxide using 1% phenolphthalein solution as an indicator. The endpoint was denoted by the appearance of pink colour. The titration was repeated thrice and the average value was recorded (Srivastava and Sanjeev, 1994). The results were expressed as a percentage using the following equation: %Acidity = Titre × 0.1 × 0.064 × 100/Weight or volume of sample. Ten grams of each replicate were homogenized with 100 mL deionized water in a 250–mL beaker by using a homogenizer (Ultra–Turrax T25, IKA, Germany). The pH of the homogenized solution was measured with a potentiometer (Micron pH 2001, Crison Instruments, S.A., Barcelona, Spain). The pH electrode was previously calibrated by using standard solutions of pH 4.0 and 7.0. The standard plate count method was followed for a total viable count of microorganisms during storage of coconut milk samples. For this, one milliliter of the coconut milk sample was aseptically measured into a sterilized McCartney bottle and 9 mL of sterile distilled water was added and shaken vigorously on a vortex mixer (XH–D, Pioway Medical Lab Equipment Co. Ltd., Jiangsu, China) which gave a homogenous suspension to be serially diluted. From homogenous suspension, 1 mL of 10−1 to 10−6 dilution was placed on a plate containing PCA agar and incubated for 24 h at 37°C. Petri dishes overloaded with bacterial colonies were avoided and colonies on each plate (having 30–300 colonies) were counted using a colony counter (Majk–Cc–30A, Guangzhou Maya Medical Equipment Co. Ltd., Guangdong, China). Finally, the total number of bacteria per gram of sample was calculated by the following equation: The coconut samples were periodically (at 7 day’s interval from processing day up to the 21st days) observed during storage for colour, fungal growth and overall status. Sensory evaluation of coconut milk was performed through a testing panel consisting of 15 members using 9 points hedonic scale. The panelists were selected from the students and employees of the Department of Food Processing and Preservation, Hajee Mohammad Danesh Science and Technology University (HSTU), Dinajpur–5200, Bangladesh, who frequently take part in such evaluation. The panelists were asked to assign the appropriate numerical score to each product for characteristics colour, flavour, taste and overall acceptability of coconut milk. The hedonic scale arranged was such that: 9 = like extremely, 8 = like very much, 7 = like moderately, 6 = like slightly, 5 = neither like or dislike, 4 = dislike slightly, 3 = dislike moderately, 2 = dislike very much and 1 = dislike extremely. Each experiment was repeated in triplicate. The obtained data were analyzed by SPSS (version 20.0). The results were expressed as mean ± standard error mean (SEM). Significant differences between the groups were assessed by one-way analysis of variance (ANOVA) test and means were separated by Duncan’s Multiple Range Test (DMRT) at the 95% confidence level.