Using seismic attributes as features for classification in feature space,in various aims such as seismic facies analysis,is conventional for the purpose of seismic interpretation.But sometimes seismic data may have no attributes or it is hard to define a small and relevant set of attributes in some applications.Therefore,employing techniques that perform facies modeling without using attributes is necessary.In this paper we present a new method for facies modeling of seismic data with missing attributes that called dissimilarity based classification.In this method,classification is based on dissimilarities and facies modeling will be done in dissimilarity space.In this space dissimilarities consider as new features instead of real features.A support vector machine as a powerful classifier was employed in both feature space(feature-based)and dissimilarity space(feature-less)for facies analysis.The proposed feature-less and feature-based classification is applied on a real seismic data from an Iranian oil field.Facies modeling using seismic attributes provide better results,but the feature-less classification outcome is also satisfactory and the facies correlation is acceptable.Indeed,the power of attributes to discriminate different facies causes to that facies analysis using attributes provide more reliable results comparing to featureless facies analysis. Using seismic attributes as features for classification in feature space, in various aims such as seismic facies analysis, is conventional for the purpose of seismic interpretation. But sometimes seismic data may have no attributes or it is hard to define a small and relevant set of attributes in some applications.Therefore, employing techniques that perform facies modeling without using attributes is necessary. This paper we present a new method for facies modeling of seismic data with missing attributes that called dissimilarity classification.In this method, the classification is based on dissimilarities and facies modeling will be done in dissimilarity space. This space dissimilarities consider as new features instead of real features. A support vector machine as a powerful classifier was employed in both feature space (feature-based) and dissimilarity space (feature-less) for facies analysis.The proposed feature-less and feature-based classification is applied on a real seismic data f rom an Iranian oil field. Facies modeling using seismic attributes provide better results, but the feature-less classification outcome is also satisfactory and the facies correlation is acceptable .dedeed, the power of attributes to discriminate different facies causes to that facies analysis using an attribute provide more reliable results comparing to featureless facies analysis.