The most efficient access to CDF variables will usually occur when the CDF has the multi-file format. The extra overhead involved with the indexing scheme used in single-file CDFs is small, so the difference may not be significant (especially if hyper reads/writes are used). The drawback to using the multi-file format is that more than one file is associated with a CDF (which may or may not be a problem for your system management).
There is a case in which the single-file format may be more efficient. If a CDF has a large number of variables (larger than the number of files that may be open at once by an application) and the variables values are accessed variable-by-variable (rather than accessing an entire variable before going to the next variable), the multi-file format may be much slower than the single-file format. This is because the CDF library will have to close one variable file and then open another as each variable value is accessed by the application (since the operating system's open file limit will be reached). If the application was to access every value for a variable before going on to the next variable, this would not occur (but it might create complications for your application).
Note that the format of a CDF can also be converted using the CDFconvert toolkit program (which creates a new CDF with the specified format). Section 3.4 describes CDFconvert.