The agribusiness of melon (Cucumis melo L.) in the Brazilian Northeast has a high representativeness in the internal and export trade scales. Eventually, productivity losses are recorded as a result of drought cycles and high evapotranspiration rates in the region. The great genetic variability of the melon is conserved in germplasm banks and collections, representing allele sources that help identify parents with characters of interest and enable the elaboration of strategies for subsequent stages of breeding programs. The objective of this research was to investigate the responses of melon accessions to water deficit during the germination, initial development and vegetative growth processes. The work was carried out in two stages, the first during germination and the second until the flowering period. The first step was divided into two trials. In both, a randomized complete block design with four replicates and 25 seeds per plot was used. In the first experiment the germination test was performed with 25 genotypes submitted to three osmotic potentials (0.0, -0.2 and -0.4 MPa). The evaluated characteristics were: germination percentage, aerial part length and radicle, aerial shoot dry mass and radicle and production efficiency index (IEP). From the genotypic means obtained with the IEP, the accessions were classified as tolerant (T), moderately tolerant (MT) and susceptible (S) to water deficit. For the second test, a representative access of each classification was selected: A-09 (T), A-16 (MT), A-02 (S), and AHK-119, a promising genotype in drought tolerance, and the hybrid 'Goldex', widely cultivated in the Brazilian Northeast. The five genotypes were evaluated for the germination speed index (IVG) in two osmotic potentials (0.0 and -0.2 MPa). Reduction of osmotic potential led to a decrease in seed germination and seedling initial development from -0.2 MPa. The development of normal seedlings was inhibited between potentials -0.2 and -0.4 MPa for all accesses tested. Goldex and A-09 presented the highest genotype means for IVG. The second stage of the work was carried out with the five genotypes evaluated for the IVG. The experimental design consisted of randomized blocks in a 5 x 3 factorial scheme (genotypes x treatments) with three replicates and fifteen plants per replicate. Treatments started 7 days after transplantation and resulted from the combination of readily available water content (AFD) and three depletion coefficients (0.35, 0.45 and 0.55). A critical stress was determined to restore moisture at each treatment. The results revealed a significant effect of the water deficit on the leaf water content (TRA) and leaf water potential (ΨWf), and on the access interaction (A) x water deficit (DH) for TRA and ΨWf at noon. The growth variables stem diameter (DC), number of leaves per plant (NFP), plant height (AP), leaf dry mass (MSF), stem dry mass (MSC) and leaf area (AF) were significantly influenced by the effect of A, the effect of DH was not significant only for AF. The root density (ρr) did not interact with the fixed or random effect. The pattern of response to drought of the melon varies according to the development phase, so the susceptibility in the early stages of growth should not be extrapolated to later stages, since drought responses demand adaptations and changes in the phenotype that become more expressive during vegetative growth