The present invention relates to a convex optimization-based high-precision positioning method in an underwater target positioning problem. The method of the present invention comprises a step of setting the to-be-calculated coordinate of a target as x=[x0,y0,z0]<T>, wherein the step comprises firstly measuring the distances ri between the target to several surrounding beacons and the coordinates ai=[xi, yi, zi]<T> corresponding to the beacons, setting the range finding errors to each beacon as Epsilon i which follow the Gaussian distribution of which the expected value is 0 and the variance is sigma i<2>, and obtaining a range finding equation of the target, etc. By carrying out the formal transformation and adding the limitation conditions on a least-square structure of an underwater target spherical intersecting positioning equation, the least-square structure is transformed into a DC structural form in a convex optimization theory, and further a convex-concave process (CCP) method can be utilized to solve, and aiming at the disadvantage that a direct CCP algorithm need to iterate an initial value in a feasible domain, a slack variable and a penalty function are added in an original optimization equation, thereby expanding the feasible domain, and broadening the limitation to the initial value. Compared with a linear least-square positioning calculation method, the convex optimization-based high-precision positioning method enables the positioning precision to be improved, and realizes the high-precision underwater target positioning.