Environmental investigations typically entail broad data gathering efforts which include field screening surveys and laboratory analyses. Although usually collected extensively, data from field screening surveys are rarely used in the actual delineation of media contamination. On the other hand, laboratory analyses, which are used in the delineation, are minimized to avoid potentially high cost. Multivariate geostatistical techniques, such as indicator cokriging, were employed to incorporate volatile organic screening and laboratory data in order to better estimate soil contamination concentrations at a underground storage tank site. In this work, the direct and cross variographies are based on a multi-scale approach. The results indicate that soil gas measurements show good correlations with laboratory data at large scales. These correlations, however, can be masked by poor correlations at micro-scale distances. Consequently, a classical direct correlation analysis between the two measured values is very likely to fail. In contrast, the presented multi-scale co-estimation procedure provides tools for a cost-effective and reliable assessment of soil contamination based on a combined use of laboratory and field screening data.