Andrej,
Bandwidth is generally considered to be a measure of the maximum amount of traffic that could travel between two points. This is a function of things such as the physical media and the provisioning by the provider of the service, that might include shaping and the like.
Wireshark can't really measure this directly, however it is possible to infer it by measuring the maximum throughput of traffic between those points, which it can measure. If you believe that a set of flows of traffic is running at close to the maximum bandwidth, the best way would be to use the IO Graph function in the Statistics menu. Then enter a "Filter" that represents the flow(s) that you think will represent the traffic when the pipe if full. For instance "ip.src="" and ip.dst=4.3.2.1". Then click on the "Graph" button and scroll until you reach a point which shows where the maximum occurs. Unfortunately the graph doesn't provide a value beyond what you can "eyeball" from the graph.
(Just remember that while a lot of protocols try to take advantage of all of the available bandwidth, it is not always the case. Some traffic type like, multimedia streaming or VoIP often have a fixed data rate so never go faster even if more available bandwidth is available. Other like a normal HTTP download will usually try to "max out" the link, however they may never reach it because of things like packet loss, delay, and so on. I know I am being a pedant, but I'm just making the point bandwidth and throughput are not the same thing)
