Markov chain Monte Carlo methods are useful in connection with inference and prediction for spatial generalized linear mixed models, where the unobserved random eects constitute a spatially correlated Gaussian random eld. We point out that so-called Langevin-type updates are useful for Metropolis-Hastings simulation of the posterior distribution of the random eects given the data. Furthermore, we discuss the use of improper priors in Bayesian analysis of spatial generalized linear mixed models with particular emphasis on the so-called Poisson-log normal model. For this and certain other models non-parametric estimation of the covariance function of the Gaussian eld is also studied. The methods are applied to various data sets including counts of weed plants on a eld. Keywords: Bayesian statistics; generalized linear mixed model; geostatistics; improper prior; Langevin-Hastings update; Markov chain Monte Carlo; Metropolis-adjusted Langevin algorithm; non-parametric covar...