The experiment was conducted in a recirculatory system in the wet laboratory of the Department of Aquaculture, Faculty of Fisheries, Bangladesh Agricultural University, Mymensingh, Bangladesh for a period of 60 days during the month of June to August, 2006. Experimental system: The experimental system consists of 20 rectangular glass aquaria (each of size 46× 41× 41 cm) containing about 65 L of water each. Altogether 12 aquaria were used for the study. All the aquaria were kept on a rack made of iron bar to facilitate better observation and accessibility. Each of the aquarium received water at the rate of 1.5 L/min. The outflow of water from the experimental aquaria drained through the standpipe. The opening of the standpipe is 1.5 cm inner and 2.0 cm outer diameter and covered by 1 mm mesh screen to prevent the escape of prawn PLs. Water was circulated through a common biological filter system under gravity before flowing into a sump tank so that all the replicate aquaria shared similar water. The biological filter drums contained net wrapped small plastic cans and small marble stones that promoted settling of wastes by increasing retention time and provided a substrate for attachment of nitrifying bacteria. At the mouth of the first filter tank a filtering device was placed to collect incoming solid wastes (uneaten food, faeces and other solid wastes) from the water. Water was pumped from the sump tank using a pump (Johnson pump, MDR-Series, Sweden) to the overhead tank for distribution into the experimental aquaria. Excess water from the overhead tank overflowed to the sump tank through a PVC pipe. A high degree of recirculation was maintained with minimal replacement of evaporation and splashing losses by fresh water make-up. An adequate level of oxygen in each aquarium was maintained through artificial aeration using an air pump (Hiblow Air Pump, SPP-100GJ-H, Techno Takatsuki Co. Ltd., Japan). Natural photoperiod of 12 h light and 12 h dark was maintained throughout the experimental period. Source of postlarvae (PLs) and acclimatization: The PL-15 of M. rosenbergii collected from Freshwater Station, Bangladesh Fisheries Research Institute (BFRI), Mymensingh were brought to Bangladesh Agricultural University in oxygenated bags. The PLs were kept in the recirculatory systems for 7 days to acclimatize with the new environment and during this time they were fed a commercial prawn nursery feed (32% protein) from Saudi-Bangla Fish Feed Ltd., Bhaluka, Mymensingh, Bangladesh. Experimental feed: A Saudi-Bangla nursery Golda feed containing 32% protein from Saudi-Bangla Fish Feed Ltd., Bhaluka, Mymensingh, Bangladesh were used for rearing the PLs in the recirculatory system. Experimental design and feeding rate: The experiment was conducted in completely randomized design. Three artificial substrates were assigned to three different treatments. Each treatment had three replicates. Treatment T1 received no substrate and considered as the control. Treatment T2 contained 2 pieces of hollow PVC pipe (15 cm long, 2 cm internal diameter) placed on the bottom of each aquarium. The PVC pipes increased the surface area and hollow areas provided more space for hiding of PLs. Treatment T3 was provided with HDPE (High Density Polyethylene) black netting (1.2 cm mesh) oriented vertically to increase available surface area in the aquarium by about 40% and attached with the aquarium wall by binder clip (Shihmark® Binder clip, SQ-155, China). Black nylon netting (1.0 cm mesh) was used as substrate in treatment T4. Piece of net material was installed vertically across the aquarium and tied with binder clips. Each net material installed in an aquarium had an area of about 1620 cm2 to increase available surface area by about 40 %. PLs were hand-counted and stocked at the rate of 75 PL (1.25 PL L-1) in each aquarium. The mean initial length and weight of PLs were 1.20 ± 0.02 cm and 27 ± 0.02 mg, respectively. PLs were fed at a rate of 20% of their body weight at the beginning and feeding rate was gradually reduced to 10% at the start of 2nd month. The total amount of feed for a day was divided in to three equal proportions and delivered at 900, 1300 and 1700 h daily. Uneaten feed, feces, and exuviae were removed by siphoning from tank bottoms immediately prior to each morning and afternoon feeding. A record of supplied feed was kept for determining the feed conversion ratio (FCR) and protein efficiency ratio (PER). Fortnightly sampling of about 25% of the total PLs was done to observe the growth of PLs and to adjust the feeding rate. The weight of PLs during each sampling was measured by a digital electronic balance (OHAUS, Model CT1200-S, Princeton, NJ, USA). Total length of M. rosenbergii PL/juvenile at the start and end of the experiment was measured from the tip of the rostrum to the tip of the telson by a centimeter scale. The sampled PLs were handled very carefully as the species is very susceptible to handling stress. Water quality parameters viz. water temperature, dissolved oxygen, pH and total ammonia were monitored weekly throughout the experimental period. Water temperature and pH were measured by a Hach pH meter (model Sension 1, Germany) and dissolved oxygen by Hach DO meter (model Sension 6, Germany) and total ammonia by a Hach kit (model DR 2010) following the standard method. At the end of the experiment all the juveniles from each aquarium were counted and batch weighed. Calculations and statistical analysis: Weight gain (g), specific growth rate (SGR % day-1), food conversion ratio (FCR), protein efficiency ratio (PER) and survival (%) of the PLs were calculated. Food conversion ratio and protein efficiency ratio were considered as apparent as no correction was made for uneaten food left (if any). The proximate composition of the feed was analyzed. Crude protein was determined by the Kjeldahl method (total N × 6.25) with a Kjeldahl System (1007 Digestion System, 1002 Distilling unit, and Titration unit, Tecator, Hoganas, Sweden) using boric acid to trap released ammonia. Lipid was determined by Soxhlet method using diethyl ether as the organic solvent. Crude fibre contents in diet samples were determined using a Fibretech system (model M1017 Hot Extractor, Tecator, Hoganas, Sweden). The nitrogen-free extract was calculated by difference. Data on weight gain, SGR, FCR, PER, and survival were analyzed using one-way ANOVA (MSTAT-C package program). Significant variation between the means was evaluated by Duncan’s new multiple range test. A significance level of P<0.05 was used.