Begoña Martínez-Cruz et al. (2012) studied the frequencies of Y-haplogroups in the Spanish and French Basque country, Gascony, Navarra, La Rioja, northern Aragon, Cantabria, and northern Castille & Leon. There are 835 samples, making it the biggest and most detailed study for the region so far.
I have attached the table with the full results. The Basque regions are:
French Basque country
ZMX : Lapurdi/Nafarroa Beherea (west)
NLA : Nafarroa Beherea (central)
SOU : Zuberoa (east)
Spanish Basque country
GUI : Gipuzkoa (east)
GSO : Southwestern Gipuzkoa
ALA : Alava (south)
BBA : Bizkaia (west)
BOC : Western Bizkaia
RON : Roncal, Nafarroa (North-Eastern Navarre)
NCO : Central/Western Nafarroa
NNO : North/Western Nafarroa
One major find is that 16.3% of the 558 Basque men tested (both Spanish and French + Navarra) belonged to R1b-L21, so far associated with the British Isles and Northwest France. It is only 1/5 of all Basque R1b, but is nevertheless significant.
In Gascony, 10.75% of men belonged to R1b-L21, while it was 8.5% in the regions of Spain bordering the Basque country. Cantabria had 0% though.
The typically Basque and Spanish R1b-M153 was found in 15.4% of the Basques, and about 8.5% in surrounding regions (again 0% in Cantabria).
The other main Iberian R1b subclade, SRY2627, is considerably lower among the Basques (6%) than the Gascons (12%) and other North Spaniarrds (11.7%). M153 and SRY2627 seem to have opposite patterns.
The most common R1b subclade was the pan-Italo-Celtic P312 (S116), which peaked at 42.3% in the Basques, and is found in 37.3% of the Gascons and 35.3% from Aragon to Cantabria.
There were only traces of the Germanic R1b-U106 (S21) : 11 samples out of 835, or 1.3%. However 6 of these samples were in Gascony (France). The 1% of I1 is consistent with that.
The Alpine Celtic and Italic R1b-U152 (S28) was tested along with two of its subclades (L2 and L20). There is no particular pattern for any of them. L2 is found at low frequencies everywhere. There were only a few samples of L20 and U152* (among the Basques all four in North Western Nafarroa). In total, U152 and suclades made up 2.3% of Basque lineages, 2.5% of Gascon lineages and 4.2% of those between Aragon and Cantabria.
The other main haplogroup of the Basques is of course I2a1a (M26, formerly I2a1), which reaches 6.6% among the Basques, with a peak at 17% at Zuberoa and 12% at Lapurdi Nafarroa, both in the French Basque country.
One surprise is the astonishingly high percentage of I2a2a (M223, formerly I2b) in Gascony (5.7%), reaching a staggering 9% in Bigorre (central-north Pyrénees) and 7% in Chalosse (north-west Pyrenees) - the highest percentage outside central Germany ! In sharp contrast, I2a2a is under 0.5% in Northeast Spain, as if the Pyrenees were a barrier to its progression. But who brought I2a2a to Gascony ? The Franks ? The equally high percentage of R1b-U106 in Chalosse points in this direction, or at least a Germanic source. Had I2a2a been Paleolithic or even had it come with the Bronze Age Celts, it would have spread to eastern Spain in greater number. This would mean that I2a2a may not have been found in Southwest Europe since the Paleolithic after all, but is a truly Germanic haplogroup like I1.
Let's note also the almost complete absence of haplogroup R1a, T and G among the Basques (only 2 G samples and 1 T sample), confirming the old discussion on the topic.
The most common subclade of E1b1b among the Basques is the Northwest African E-M81 (1.25%) followed by the North African E-V65 (1%). Other subclades are just found at trace frequencies, even E-V13 (only 1 sample).
On the J front, the Basques have a reasonable amount of J2a (1.6%), but very little J2b (0.35%), even less than J1 (0.55%).
The absence of E-V13, T and G2a would suggest that these three haplogroups migrated together, presumably during the Neolithic. On the other hand, the presence of J2 alongside a similar frequency of R1b-U152 would hint at a Gallo-Roman origin of these two haplogroups (I say Gallo-Roman as it could be from Gaul or Italy, the two regions being closely tied at least since the Bronze Age).