HI, I'd thought that I'd add my experience on this. We experienced the same issue with the log reader agent in Sql 2000, caused by a now outage over the weekend. Instead of amending the -readbatchsize to 1, I amend from 500 to 100, which also worked. I suppose it depends on the severity of the backlog. So you don't need to reduce all the way down to single batch sizes.