Accurate calculation of iodine behavior in the containment is very important in determining the potential radioactive release to the environment in light water reactor severe accidents (SAs). Of particular significance is the behavior of gas phase iodine, particularly organic iodine, which is difficult to remove by filtration, e.g., in containment venting systems. Iodine behavior is closely linked with the containment thermal hydraulics, which have a major influence on the distribution of iodine throughout the containment atmosphere and sump. In the European 7th Framework SARNET project, European Commission (EC) cofunded from 2007 to 2013, SA modeling code capability was assessed through two integral benchmarks. In the first, the basis was the German THAI Iod-11/12 tests, where molecular iodine transport with atmospheric flows and iodine interactions with steel surfaces were emphasized. In the second, data from the international Phébus FPT3 test were used, where all aspects of SAs were studied from core degradation, fission product (FP) release, circuit transport/deposition, and containment behavior using realistic FP sources. Thermal hydraulics in the containment were simpler, being well-mixed, and radiolytic interactions of iodine, e.g., with painted surfaces, were studied. These interactions may be an important source of organic iodine in the containment atmosphere. The two benchmarks are thus complementary. In the FPT3 exercise, the calculations could predict the containment thermal hydraulic conditions fairly well. For the more detailed data from THAI, differences were noted for atmospheric flows and relative humidities, outside experimental uncertainties, affecting iodine behavior. The FPT3 iodine results themselves showed a spread in calculated results outside data uncertainties, indicating the need for model improvements in this area, e.g., for radiolytic interaction of iodine with paint. Experimental programs to generate the necessary data needed for code improvement have been recently completed, e.g., in the OECD/THAI, THAI2, BIP, and BIP2 projects, or are in progress, in OECD/STEM and EC/PASSAM. When model improvements have been made, repeat benchmarks are planned to check progress toward code convergence with experimental data, e.g., under the aegis of the new NUGENIA association of which SARNET now forms a part.