Use exceptions for exceptional circumstances, not for simple logic errors that can be caught easily using an if statement.
The try block contains the code that is executed in normal circumstances. The catch block contains the code that is executed in exceptional circumstances. Exception handling separates error-handling code from normal programming tasks, thus making programs easier to read and to modify. Be aware, however, that exception handling usually requires more time and resources, because it requires instantiating a new exception object, rolling back the call stack, and propagating the exception through the chain of functions invoked to search for the handler.
An exception occurs in a function. If you want the exception to be processed by its caller, you should throw it. If you can handle the exception in the function where it occurs, there is no need to throw or use exceptions.
In general, common exceptions that may occur in multiple classes in a project are candidates for exception classes. Simple errors that may occur in individual functions are best handled locally without throwing exceptions.
Exception handling is for dealing with unexpected error conditions. Do not use a try- catch block to deal with simple, expected situations. Which situations are exceptional and which are expected is sometimes difficult to decide. The point is not to abuse exception handling as a way to deal with a simple logic test.
A general paradigm for exception handling is that you declare to throw an exception in a function as shown in (a) below, and use the function in a try-catch block as shown in (b).
Source: Liang Y. Daniel (2013), Introduction to programming with C++, Pearson; 3rd edition.