We perform Monte Carlo simulations to study the two dimensional random-bond XY model on a square lattice. Two kinds of bond randomness with the coupling coefficient obeying the Gaussian or uniform distribution are discussed. It is shown that the two kinds of disorder lead to similar thermodynamic behaviors if their variances take the same value. This result implies that the variance can be chosen as a characteristic parameter to evaluate the strength of the randomness. In addition, the Berezinskii-Kosterlitz-Thouless transition temperature decreases as the variance increases and the transition can even be destroyed as long as the disorder is strong enough.PACS numbers: 75.50. Lk, 75.40.Mg, 05.70.Fh, 05.70.Jk The Berezinskii-Kosterlitz-Thouless (BKT) transition in two dimensional system has been extensively studied for decades since the discovery of the exotic quasi-longrange order formed by the binding of vortex-antivortex pairs. [1,2] The simplest model to demonstrate the BKT transition is the so-called 2-Dimensional (2D) XY model and the Hamiltonian takes the formwheredonates spin at site i. J ij is the coupling coefficient of the two nearestneighboring sites, i and j, and θ ij = θ i − θ j is the phase difference between the two sites. Typically, the transition in the 2D XY model is characterized by low temperature power-law decay of a two-point correlation function which gives rise to divergent susceptibility, [3] measurable finite-size-induced magnetization with universal magnetic exponent[4] and a discontinuous jump to zero of the helicity modulus. [5] In experiments, the BKT transition has been confirmed in various real systems such as 4 He films,[6] Josephson-junction arrays [6] and planar lattice of Bose-Einstein condensates. [8] Due to the presence of defects and distortions, real systems are always imperfect and usually subject to certain disorder effects. Therefore it is of interest to study how the imperfection affects on the BKT transition. For the 2D XY model, the imperfection can be demonstrated by two parameters, J ij and θ ij . J ij (or θ ij ) may be governed by a random distribution P (J ij ) (or P (θ ij )). It means that J ij (or θ ij ) takes the values subject to a probability distribution, which is called the bond randomness (or phase randomness).Rubinstein, Shraiman, and Nelson studied 2D XY ferromagnets with random Dzyaloshinskii-Moriya interactions and derived a model with both J ij and θ ij randomness.[9] They showed that the spatial variation in † Corresponding author: qgu@ustb.edu.cn J ij might be irrelevant at long wavelengths. For this reason, less attention was paid on this type of disorder. However, the case that J ij obeys a discrete probability distribution has still been intensively studied. A simple choice of [10][11][12] which means each bond might be vacant with probability 1 − p. This model is often referred to as the bond diluted model.In fact, the continuous bond randomness case that J ij obeys a continuous probability distribution should not be neglected. Korshunov arg...