This paper compares continuous surfaces obtained by the interpolation of standard penetration test values (Nspt) by depth using deterministic and geostatistical models. The evaluation of the interpolation methods was based on the percent areas of the compactness classes by depth obtained in each method and on the percent errors calculated from the differences between the measured and interpolated Nspt. The Nspt values were obtained from 228 standard penetration tests (SPT) in situ investigations carried out in quaternary sandy marine sediments located along the Brazilian northeast coast. The boreholes are well distributed in an area of 10.9 km2, covering six districts of Joao Pessoa City (Paraiba State, Brazil). The interpolations and analyses were conducted using geographic information system and spreadsheet software. The interpolations of Nspt values were made at depths of 1, 3, 6 and 15 m. The interpolated Nspt values obtained for each depth were classified into five soil compactness classes. The general boundaries of the compactness classes and their percent areas by depth were similar for both methods with some localized differences. Both methods presented high average percent errors and standard deviations, with a slight advantage to the geostatistical model (ordinary kriging). The minimum average percent error and standard deviation occurred at a depth of 3 m for both methods, with a slight advantage to the deterministic method (spline modified). Both methods also indicated a peculiar high soil resistance of the marine sediments at a depth of 6 m and significant areas of loose soils at a depth of 15 m in one of the districts analyzed. Both methods showed similar potential for practical applications for the initial phases of a project, not ruling out more specific SPT tests, but can optimize them.