Optimization analysis of tandem queues attended by a moving server with holding and switching costs are the main concern in this research. First, as basic models, two-stage tandem queues attended by a moving server are analyzed to find the optimal policies in the first and the second stages in a two-stage tandem queue. It is shown that the optimal policy in the second stage which minimizes the total discounted and long run average holding and switching costs is a greedy and exhaustive policy. Considering greedy and exhaustive policy for the second stage, three different service policies: (i) static, (ii) semi-dynamic (gated-limited) and (iii) dynamic (double-threshold) policies are defined for the first stage of a two-stage tandem queue, an usdM/G\sb1 - G\sb2/1usd queue, to respectively deal with situations that (i) no information, (ii) partial information, or (iii) complete information about the number of customers in the system are available. To obtain the optimal static, gated-limited and double-threshold policies, three different models are developed and it is shown that the optimal gated-limited policy is almost independent of the arrival rate if switchover times are zero. For N-stage tandem queue usdM/G\sb1 - G\sb2 -{\cdots}-G\sb{N}/1,usd greedy and exhaustive policies downstream of stage 2 are defined and then three different models are developed to find the optimal static, gated-limited and double-threshold policies which minimize the total average holding and switching costs in these queues. Also, optimal policies in N-stage tandem queues with batch arrivals, usdG\sp{(x)}/G\sb1-G\sb2-{\cdots}-G\sb{N}/1usd queues, are studied through a new operational tool which reveals the properties of greedy and exhaustive policies and establishes two efficient and accurate heuristic algorithms to obtain the optimal policy. Finally, U-shaped lines which are actually tandem queues attended by moving servers are reviewed and a general definition and classification scheme are presented which organize the previous research and lead to the potential for further research on these lines. Then, by decomposing U-shaped lines into tandem queues attended by a moving server, the effects of switching costs and switchover times on some types of these lines are analyzed.