In the Pawlak model of conflict analysis, a rating of −1, 0, and +1 indicates that an agent is negative, neutral, and positive towards an issue. One defines measures of alliance and conflict of agents based on their three-valued ratings on a set of issues. Existing studies make some restrictive assumptions. One assumption is that a single distance function determines alliance, conflict, and neutrality. Another assumption is that all issues are equally important. The third assumption is that different pairs, such as $(+1,+1)$, $(0,0)$, and $(-1,-1)$, contribute equally to alliance and conflict. Intuitively, different issues may have different importances and a $(+1,+1)$ agreement may be stronger than a $(0,0)$ agreement. To address these weaknesses, in this paper we use a pair of an alliance measure and a conflict measure, rather than a single distance measure. We propose and study a new class of additive measures of alliance and conflict by assigning different weights to different issues and assigning different values to different pairs of ratings. Existing measures are special cases of the new measures. By using the new measures, we construct a general model of three-way conflict analysis. Finally, we give an application of the new alliance and conflict measures in making the development plan of Gansu Province in China.