It depends on the degree of efficency required and the size of data one is working with.
When thinking of Perl in the mind set of it original goal of the language the <> method makes prefect sense (easy things easy). You read in a small text file and either report on it or make a small change, but when you look at reading in a 56MB log file that someone forgot to put newline characters on (that really happened), then doing it using the read method makes more sense. (hard things possible)