This study considers the optical critical dimension (OCD) measurement for parameter estimation with multiple observation vectors in semiconductor manufacturing. The main objective of this study is to provide not only the estimated CD values but also the amount of uncertainty of the values. We address the problem of estimating parameter vectors with respect to a Bayesian perspective. Since it is difficult to directly derive the posterior distribution of the parameter vectors, we apply MCMC algorithms to sample parameter vectors and estimate the posterior distribution. In a case study problem, we consider a 2-dimensional high-aspect-ratio structure of a wafer in semiconductor manufacturing and we numerically compare three existing MCMC algorithms, (i) Metropolis-Hastings (MH), (ii) Multiple-Try Metropolis (MTM), and (iii) Parallel-Multiple-Try Metropolis (P-MTM), regarding accuracy and efficiency. The experimental results show that the P-MTM algorithm achieves the highest accuracy and efficiency compared to the MH and MTM algorithms.