Abstract: The degree correlation of 50 urban street networks in their Stroke-based dual representation is measured by both Newman's assortativity coefficient and Litvak-Hofstad's assortativity coefficient. In addition, because of modified linear unit problem, the two assortativity coefficients are calculated and analyzed over different Stroke sets of a street network. Finally the relationship between degree correlation and robustness is investigated. It is found that: (1) An overwhelming majority of urban street netwos in their dual representation are disassortative and uncorrelated in terms of the degree, although the two assortativity coefficients can be inconsistent sometimes. (2) For a given street network, different Stroke sets have slight influence on the Newman's assortativity coefficient and no influence on the Litvak-Hofstad's assortativity coefficient. In general, the influence of different Stroke sets on the measures of degree correlation is minor and limited. (3) When measured by Litvak-Hofstad's assortativity coefficient, the degree correlation of urban road street network is positively correlated with its robustness.