designing primers in an intron allow you to sequence the entire exon as you cannot sequence under the primer. Also genes often have similar sequences ( functional motifs) in their exons so designing a primer in the intron is often more specific than exonic primers especially where the is a gene family of similar genes like HOX genes. This concept does not apply when the intronic/intragenic sequence is a repeat sequence which should be avoided
Adding to Paul's answers, in some cases the exon sequences are more conserve than the intron ones. Therefore, when you design primers for amplifying part of genes using PCR and if you use genomic DNA as PCR template, I would suggest you use part of the "conserved sequences" in the exon regions instead of introns.
There will be a possibility of mispriming because of probable extensive changes in the intronic sequences since they are less conserve than the exons (well except for intron splice junctions, which are very conserved in most genes). Moreover, if you use cDNA as PCR template - you will not be able to use intronic primers for PCR.
Primers are designed from the intronic part (60-80 nucleotide) up stream, for forward and the same for the reverse, because we want to sequence the exonic part. If the primer is designed close to exon then some part of the exon might not be covered (sequenced) properly.
we also want to make sure that splice site variants are not skipped....
If you are designing primers for the amplification of cDNAs for RT PCR then the sequence of the primers has to be based on adjacent exons (like exon 1 and exon 2) and if it is for a gene in the genomic region then it is better to have primers designed from exons and adjacent introns.